A Linguistic Commentary on Livius Andronicus (Studies in Classics)

RT8992_FM.qxd

6/25/04

12:20 PM

Page i

STUDIES

IN

CLASSICS

Edited by

Dirk Obbink

Andrew Dyck

Oxford University

The University of California, Los Angeles

A ROUTLEDGE SERIES

RT8992_FM.qxd

6/25/04

STUDIES

12:20 PM

IN

DIRK OBBINK

Page ii

CLASSICS

AND

ANDREW DYCK, General Editors

SINGULAR DEDICATIONS Founders and Innovators of Private Cults in Classical Greece Andrea L. Purvis EMPEDOCLES An Interpretation Simon Trépanier APHRODITE AND EROS The Development of Greek Erotic Mythology Barbara Breitenberger RHETORIC IN CICERO’S PRO BALBO An Interpretation Kimberly Anne Barber “FOR SALVATION’S SAKE” Provincial Loyalty, Personal Religion, and Epigraphic Production in the Roman and Late Antique Near East Jason Moralee AMBITIOSA MORS Suicide and Self in Roman Thought and Literature Timothy D. Hill

RT8992_FM.qxd

6/25/04

12:20 PM

Page iii

A LINGUISTIC COMMENTARY ON LIVIUS ANDRONICUS

Ivy Livingston

ROUTLEDGE New York & London

RT_Disclaimer_6x9.fm Page 1 Tuesday, August 17, 2004 11:44 AM

Published in 2004 by Routledge Taylor & Francis Group 270 Madison Avenue New York, NY 10016 www.routledge-ny.com Published in Great Britain by Routledge Taylor & Francis Group 2 Park Square Milton Park, Abingdon Oxon OX14 4RN www.routledge.co.uk Copyright © 2004 by Taylor & Francis Group, a Division of T&F Informa. Routledge is an imprint of the Taylor & Francis Group. Printed in the United States of America on acid-free paper. All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. 10

9

8

7

6

5

4

3

2

1

Catalog record is available from the Library of Congress

Parentibus et Praeceptoribus

Contents Series Editors’ Foreword

ix

Preface by Michael Weiss

xi

Acknowledgments Introduction

xxi 1

Commentary* Section One: The Odyssey 2 f i¯lie , 5 Laertie

5

10 Morta

7

15 nequi¯nont , 36 inserinuntur

13

25, 33, 34 topper

17

30 Mon e¯t a¯s

23

33 hom o¯n e¯s

31

39 g a¯vi¯si¯

37

40 pulla

39

41 mandisset

41

46 dextr a¯bus

43

Section Two: Tragedies 4 praeda

47

20 op e¯s, opitul a¯

57

29 praest o¯l a¯r a¯s

61

38 nefrendem, lacteam inmulgens opem

67

Conclusion

71

Appendix

73

Bibliography

83

Index

85

* The numbering of the fragments follows Warmington, Remains of Old Latin.

vii

Series Editors’ Foreword

Studies in Classics aims to bring high-quality work by emerging scholars to the attention of a wider audience. Emphasizing the study of classical literature and history, these volumes contribute to the theoretical understanding of human culture and society over time. This series will offer an array of approaches to the study of Greek and Latin (including medieval and Neolatin), authors and their reception, canons, transmission of texts, ideas, religion, history of scholarship, narrative, and the nature of evidence. While the focus is on Mediterranean cultures of the Greco-Roman era, perspectives from other areas, cultural backgrounds, and eras are to be included as important means to the reconstruction of fragmentary evidence and the exploration of models. The series will reflect upon the role classical studies has played in humanistic endeavors from antiquity to the present, and explore select ways in which the discipline can bring both traditional scholarly tools and the experience of modernity to bear on questions and texts of enduring importance. Dirk Obbink, Oxford University Andrew Dyck, The University of California, Los Angeles

ix

Preface Michael Weiss Cornell University

The facts and not-quite-facts about Livius Andronicus, his life and works are available in any number of reliable handbooks, but for convenience I will rehearse them here.1

Date According to Cicero (Brutus 72
/73), who names Atticus and antiqui commentarii as his sources, Livius first presented a play in the consulship of C. Claudius Ap. f. Centho and M. Sempronius C. f. Tuditanus (240 BCE). Cicero sticks to the same date at Tusc. 1.32 and Cato 50.3 And indeed this date is followed by most later scholars both ancient and modern.4 But Cicero also reports that Livius’ dates were a subject of debate because Accius claimed that Livius was captured from Tarentum in the fifth consulship of Quintus Maxumus (209 BCE). Further Accius reported that he produced a play in 197 BCE during the consulship of C. Cornelius and Q. Minucius at the Ludi Iuventatis , which M. Livius Salinator had vowed to establish at the battle of Sena in 207 BCE.5 This later chronology seems to have been followed by Suetonius since it makes an appearance in Jerome’s chronology, but it has found few modern supporters.6 Cicero claims that Accius’ chronology is wrong because it would make Livius, supposedly the first to put on a play, younger than Plautus and Naevius. It is obvious that Cicero’s argument is a petitio principii since we do not know what dates Accius would have attributed to Plautus and Naevius. Whatever the exact dates of Livius Andronicus, it is clear that the native tradition regarded him as the initiator of Latin literary production.7 xi

xii

A Linguistic Commentary on Livius Andronicus

Name In the ancient sources Livius Andronicus is generally referred to as just Livius. The cognomen Andronicus appears for the first time in Quintilian, although Festus p. 446.32 L following Verrius Flaccus allows us to trace the second name back at least to the Augustan period.8 Given standard Roman naming practice for freedmen it is plausible that Andronicus might represent the author’s original Greek name and Livius the gentilicium of his patron and quondam owner, reported in some sources to have been M. Livius Salinator.9 The praenomen is given as L. by Aulus Gellius and Cassiodorus but as T. by Jerome, which presumably is the result of confusion with the much more famous T. Livius.10

Origin Cicero reports that Accius wrote that Livius was captured from Tarentum.11 Although this fact is embedded in a chronology that is not usually found credible, many scholars have accepted it. However, while it is true that Roman influence and hegemony had begun in Tarentum by the end of the Pyrrhic war in 272, the city was not captured nor its people enslaved. So while the Tarentine origin of Livius can be easily fitted into the high chronology, it is harder to understand how Livius could have come to Rome as a war-captive. Suetonius (De Grammaticis et Rhetoribus 1) refers to him along with Ennius as semigraecus . It is, of course, evident from his work that he was familiar with Greek and this may be all we can say with certainty.

Livius’ Status and Employment Suetonius reports that like Ennius Livius is known to have taught both Greek and Latin privately and publicly at Rome. Suetonius’ claim that they merely clarified the meaning of Greek authors and gave readings from their own Latin compositions is presumably an inference based on grammatical practice in his day.12 Jerome, relying on Suetonius’ De Poetis, adds another detail about Livius’ teaching duties and servile status: Livius Salinator freed the poet Livius, who had been his children’s instructor, ob ingenii meritum in the year 188/7. Since the date given agrees with Accius’ chronology it is conceivable that Suetonius drew this information directly or indirectly from that author.13 The association of Livius Andronicus and Livius Salinator in Jerome seems to confirm this since the Accius passage dates a play of Livius to the Ludi Iuventatis of 197 BCE, which had been vowed by Livius Salinator at the battle of Sena in 207 BCE. It is conceivable that the two dates for Livius Andronicus’ floruit result from a confusion of Livii Salinatores. Most scholars believe that the Salinator who freed Livius was the M. Livius Salinator, recorded as

Preface

xiii

magister Xvir. of the fictive Ludi Saeculares of 236 BCE in the Fasti Capitolini .14 This Livius Salinator was the father of the famous M. Livius Salinator consul and dictator, and the grandfather of C. Livius Salinator consul of 188.15 It is conceivable that Accius’s dating, if indeed erroneous, results from a confusion of the decemvir with the homonymous consul of 219 and 207. Of course it is also just possible that the entire association with the Livii Salinatores has been fabricated on the basis of the gentilic Livius and that the association of M. Livius Salinator the elder with the Ludi Saeculares , celebrated in Rome at the place in the Campus Martius called Tarentum/Terentum, led to the connection of Livius with the homonymous city of Magna Graecia.16

Literary Career and Works To Livius are attributed dramatic works, a hymn and a translation of the Odyssey. The first play was staged in 240 BCE, the year after the end of the first Punic war. Livy in his famous account of the origins of drama at Rome (7.2.8) recounts how Livius was the first to create a fabula with a plot (argumentum ). Further Livy relates that upon straining his voice Livius, who was himself an actor, left the singing to a professional and gesticulated in a livelier fashion since he was not bothered with having to sing himself. From that time on actors limited themselves to dialogue portions only (diverbia ).17 The following tragic titles are known: Achilles, Aegisthus, Aiax Mastigophorus, Andromeda, Antiopa, Danae, Equos Troianus, Hermiona, Tereus.18 Festus preserves six fragments that are attributed to comedies. Two titles, Gladiolus and Ludius , are certain but a third, Virgo , is probably corrupt. Livius was also said to have composed a hymn for a chorus of twenty-seven maidens for an expiatory rite of 207.19 For the success of this hymn and the rite it formed a part of, a collegium scribarum histrionumque was established in Livius’ honor in the temple of Minerva on the Aventine, where actors and writers could gather and offer gifts.20 Most famously of all, Livius translated the Odyssey into Saturnians.21 At what stage in his career and for what reasons he undertook this task is unknown. The frequently repeated idea that Livius’ Odysseia 22 was intended as a school-text is unsupported by any direct ancient testimony. Of course the well-known passage Ep. 2.1.69
/71 shows that some of Livius did become a school-text, much to Horace’s regret.23

Meter While the dramatic works are written in meters adapted from Greek, the iambic senarius and the trochaic septenarius , the Odyssia is written in the apparently native

xiv

A Linguistic Commentary on Livius Andronicus

and famously problematic Saturnian. It is unlikely that Livius Andronicus himself was solely responsible for the adaptation of Greek dramatic meters. As Fraenkel has pointed out, the innovations distinguishing even Livius’ Latin senarii and septenarii from their Greek models are not likely to be the work of just a single lifetime. Furthermore, the so-called versus quadratus , itself based on Greek models, appears to have been long adapted to popular non-literary use.24 Finally, it has recently been claimed that the South Picene metrical inscriptions dating from the 5th century BCE make use of Greek-derived meters.25 On the other hand, no serious attempt has been made in modern times to trace the Saturnian to Greek sources and the principles of its scansion remain controversial to the present day.26 It is, however, common to Latin and the Sabellic languages, but whether this commonality is to be explained as the result of shared inheritance from Proto-Italic or metrical diffusion within the Italic cultural koine´ is unclear.27

The Work in Question The majority of scholarship on Livius Andronicus has traditionally been strictly philological or literary-historical. In particular, much work has focussed on the relationship between Livius’ translations and their Greek originals.28 The one full-scale study devoted specifically to the language of Livius is the unpublished University of Minnesota dissertation completed in 1975 by Gabriele Erasmi, Studies on the Language of Livius Andronicus . This gigantic work, a storehouse of useful information, includes a descriptive grammar of the fragments and detailed studies of the Livian onomasticon. The present monograph, however, has a different focus. Since the meager fragments of Livius represent our earliest*/ or nearly earliest */literary Latin they provide valuable evidence for the language before the heavy hand of Latin literary tradition led to morphological and syntactic streamlining. By interrogating the fragments from the point of view of historical and Indo-European linguistics one may hope to gain insight into the pre-history of Latin and the initial development of the Latin literary tradition. In the 1997 dissertation of Ivy Livingston, Livian forms are the starting points for far-reaching discussions of Latin historical phonology and morphology. Among the most interesting results are the elucidation of the prehistory of the vocatives of -ii' o stems, the class of deverbal nouns in -e¯tum and -e¯ta , the prehistory of the paradigm of homo , and the etymology and morphology of praeda and praestolari . In an appendix a new and plausible theory for the origin of the Latin adjective in -ulentus is offered.

Preface

xv

Addenda and Corrigenda Pg. 3 The Umbrian forms heris and heri are not the best examples for the fourth conjugation in Umbrian since they could well be optative forms, either *heri¯- or heri'i¯-. A clearer example in Umbrian would be amparitu ‘stand’ (tr.). The analogically remade vocative in *-ii' e is reflected also in Oscan statie (CM 18 /Vett. 110).




Pg. 12 /13 The avoidance of monosyllables as a motivation for innovation has been studied in detail by Lo¨fstedt 1956:35
/62.

Pg. 36n3 Ved. a´vati is perhaps better taken with Latin iuvo not aveo . See LIV2:243.

Pg. 39 Warmington Od. 41 is one of the four hexameters attributed to Livius Andronicus. Presumably they date to a later post-Ennian reworking of the Odyssia . For other possible reconstructions of the root behind mando see LIV2:442.




Pg. 53 /54 The connection of opus with epulum is dubious. There is no reason to think that an epulum was originally only a religious banquet nor for that matter that opus and its cognates have any particular sacral/cultic connection. Vedic a´¯pas is a hapax of uncertain meaning. I would prefer to derive epulum from the root *h1 ep - ‘take.’ For the semantics cf. Serbian and Croatian rucˇak ‘meal’ B/– rucˇati ‘take (a meal).’ If one does not connect epulum and opus then there is no formal difficulty in connecting both opus ops with PIE *h3 ep -. The semantics, however, require further elucidation.

Concordance of Fragments Treated Warmington Warmington Warmington Warmington

Od . Od . Od. Od.

2 /Bla¨nsdorf 2 5 /Bla¨nsdorf 4 10 /Bla¨nsdorf 23 15 /Bla¨nsdorf 11

xvi

A Linguistic Commentary on Livius Andronicus

Warmington Warmington Warmington Warmington Warmington Warmington Warmington Warmington Warmington

Od. Od . Od . Od. Od. Od . Od. Od. Od.

23
/26 /Bla¨nsdorf 18 30 /Bla¨nsdorf 21 33 /Bla¨nsdorf 25 34 /Bla¨nsdorf 24 35
/36 /Bla¨nsdorf 34 39 /Bla¨nsdorf 22 40 /Bla¨nsdorf 27 41 /Bla¨nsdorf 39 46 /Bla¨nsdorf 29

Abbreviations DNP/ DKP / GLK / LIV2 / OCD3 / RE/

Der Neue Pauly Enkyklopa¨die der Antike edited by Hubert Cancik and Helmuth Schneider, Stuttgart Weimar, 1996. Der Kleine Pauly Lexikon der Antike edited by Konrat Ziegler and Walther Sontheimer, Munich, 1975. Grammatici latini ex recensione Henrici Keilii, Leipzig, 1857
/1880. Lexikon der indogermanischen Verben , second edition, edited by Helmut Rix, Wiesbaden, 2001. Oxford Classical Dictionary , third edition, edited by Simon Hornblower and Antony Spawforth, Oxford and New York, 1996. Paulys Real-encyclopa¨die der classischen Altertumswissenschaft . Neue bearbeitung, edited by G. Wissowa, Stuttgart, 1894
/.

NOTES 1. See especially Suerbaum 2002:93
/104, OCD3:876
/877 (Jocelyn); DNP:7:373
/ 378 (Suerbaum) DKP:3:692
/695 (Mariotti); RE Suppl. V:598
/607 (Fraenkel). Thanks to Rip Cohen, Michelle Kwintner and Alan Nussbaum for advice and commentary. 2. Annis fere CCCCCX post Romam conditam Livius fabulam dedit C. Claudio, Caeci filio, M.Tuditano consulibus anno ante natum Ennium, qui fuit maior natu quam Plautus et Naevius. 3. Vidi (i.e. Cato) etiam senem Livium; qui cum sex annis ante quam ego natus sum, fabulam docuisset Centone Tuditanoque consulibus, usque in aduluscentiam meam processit aetate . 4. Gellius 17.21.42 probably following Varro. 5. However, Livy 36.36.4 dates these games to 191. 6. Hier. chron. a. Abr. 1829/30/188/187 T. Livius tragoediarum scriptor clarus habetur, qui ob ingenii meritum a Livio Salinatore, cuius liberos erudiebat, libertate donatus est . For the Suetonian origin of this passage see Kaster 1995:49. Valerius Antias apud Livy 36.36.4 also may have been following Accius’ chronology since he

Preface

7.

8.

9.

10.

11.

12. 13. 14. 15. 16. 17. 18. 19.

20. 21. 22. 23.

xvii dates the first theatrical performances to the Megalesia of 191 BCE, the same year as the celebration of Livius Salinator’s Ludi Iuventatis according to Livy. The most notable modern defenders of Accius are Mattingly 1957 and 1971 and Marconi 1966. At least as the founder of Roman theater: Horace Ep . 2.1.62 with Schol. Hor. G qui primus comoedias scripsit ; Livy 7.2.8
/10; Aulus Gellius 17.21 primus omnium L. Livius poeta fabulas docere Romae coepit ; Val. Max. 2.4.4; Gloss. Lat. Ansil. p. 128, s.v. comoedia ; Diom. GLK 1.489.7; Euanthius com . 4.3. As first poet: Quinitilian Inst. 10.2.7 nihil in poetis supra Livium Andronicum . It is also found in Aulus Gellius 18.9.5, Priscian GLK II 208, 301, 305.8, 321.6, the Schol. Hor. G, Gloss. Lat. Ansil., Diom. and Gloss. Lat. Ansil. passages quoted in fn. 7 and the Old Church Slavonic translation of a work attributed to Rhetorius. However, it is noteworthy that inscriptions prior to 100 BCE do not record the cognomen of a freedman. See Gordon 1935. If the cognomen Andronicus has any validity in this case it must have been preserved in some other fashion. Aulus Gellius 17.21 quoted above in fn. 7. Cassiodorus chron. II 128 M. 316: C. Manlius et Q. Valerius. His consulibus (239 BCE) ludis Romanis primum tragoedia et comoedia a L. Livio ad scaenam data. ' nikow is a common name in the Hellenistic period, but for what it’s worth it Andro is attested for Tarentum by Fouilles de Delphes III (4) 427 III.1 (205
/203 BCE). Note, by the way, that Accius says that Livius was captured from Tarentum. That he was a native of the same place is merely an inference. See Kaster 1995:54. Note, however, that Beare 1940 has questioned Livius’ servile status. He suggests that Livius’ servile status may be an inference of imperial date. See Broughton 1951
/2:1.223. See Broughton 1951
/2:1.236, 294, 365. A different reconstruction in Palmer 1974:96 who amalgamates the decemvir and the consul. The controversies of this passage are many and the bibliography is enormous. See the sampling offered by Oakley 1998:40. See Suerbaum 2002:98 for more details about the probable Greek sources for these plays. Livy 27.36.3
/4; Festus 446.26 L s.v. scribas . Suerbaum 2002:99 suggests that the statement of Serv. auct. ad Aen. 4.37 Livius autem Andronicus refert eos (sc. Afros) de Romanis saepius triumphasse suasque porticus Romanis spoliis adornasse may be derived from this hymn. Festus p. 446, 448 L. On this collegium and its relationship to the collegium poetarum mentioned by Valerius Maximus 3.7.11 see Horsfall 1976. Suerbaum 1992 argues that the translation must have been much abbreviated. This spelling is the only one authoritatively transmitted. The frequent spelling Odusia is a restored archaic spelling not directly attested. Non equidem insector delendaue carmina Liui/esse reor, memini quae plagosum mihi paruo/Orbilium dictare . An Old Church Slavonic translation of a lost Greek work attributed to Rhetorius mentions Livius Andronicus in a hemerological context. See Benesˇevicˇ 1925. Pingree 1977:223 doubts the attribution of the OCS text to Rhetorius.

xviii

A Linguistic Commentary on Livius Andronicus

24. See Fraenkel 1927 and Gerick 1996. See also Rix 1989 for arguments for the antiquity of iambic shortening. 25. Eichner 1990. 26. The most recent theory is put forth in Parsons 1999. 27. See Poccetti 1983. A probable Etruscan example of the versus quadratus is found on the 4th century BCE mirror from Volterra: eca sren tva/ i xnac hercle / unial clan / u[u]ra sce . ‘This image shows how Hercules became the son of Uni.’ See Gerick 1996:19. 28. In particular Fra¨nkel 1932; Mariotti 1952; Traina 1953; Verrusio 1942; Broccia 1974; Bu¨chner 1979; Sheets 1981; Kearns 1990.

REFERENCES Benesˇevicˇ 1925: W. Benesˇevicˇ, ‘‘Spuren des Werke des A¨gypter Rhetorios, des Livius ¨ bersetzung,’’ Byzantinische Zeitschrift , Andronicus und des Ovidius in altslavischer U 25 (1925) 310
/312. Beare 1940: William Beare, ‘‘When Did Livius Andronicus Come to Rome?’’ Classical Quarterly , 34 (1940) 11
/19. Bla¨nsdorf 1995: Ju¨rgen Bla¨nsdorf, Fragmenta poetarum latinorum epicorum et lyricorum praeter Ennium et Lucilium post W. Morel novis curis adhibitis edidit Carolus Buechner editionem tertiam auctam curavit Ju¨rgen Bla¨nsdorf, Stuttgart and Leipzig, 1995. Broccia 1974: Giuseppe Broccia, Ricerche su Livio Andronico epico , Padua, 1974. Broughton 1951
/2: T. Robert S. Broughton, The Magistrates of the Roman Republic , with the collaboration of Marcia L. Patterson. New York, 1951
/52. ¨ bersetzung Bu¨chner 1979: Karl Bu¨chner, ‘‘Livius Andronicus und die erste ku¨nstlerische U der europa¨ischen Kultur,’’ Symbolae Osloenses , 54 (1979) 37
/70. Eichner 1990: Heiner Eichner, ‘‘Ein Heldendenkmal der Sabiner mit trocha¨ischen Epigramm eines pikenischen Plautus des fu¨nften Jahrhunderts v. Chr.,’’ Sprache 34 (1988
/1990), 198
/206. Erasmi 1975: Gabriele Erasmi, Studies on the Language of Livius Andronicus, unpublished dissertation, University of Minnesota, 1975. Fra¨nkel 1932: Hermann Fra¨nkel, ‘‘Griechische Bildung in altro¨mischen Epen,’’ Hermes, 67 (1932) 598
/607. Fraenkel 1927: Eduard Fraenkel, ‘‘Die Vorgeschichte des versus quadratus ,’’ Hermes, 62 (1927) 357
/370. Reprinted in Kleine Beitra¨ge zur klassischen philologie , Vol. 2, Rome, 1964, 11
/24. Gerick 1996: Thomas Gerick, Der versus quadratus bei Plautus und seine volkstu¨mliche Tradition, Tu¨bingen, 1996. Gordon 1935: Arthur E. Gordon, Ephigraphica, I. On the First Appearance of the Cognomen in Latin Inscriptions of Freedmen, University of California Publications in Classical Archaeology 1.4, 1935. Horsfall 1976: Nicholas Horsfall ‘‘The Collegium Poetarum ,’’ Bulletin of the Institute of Classical Studies, 23 (1976) 79
/95. Kaster 1995: Robert A. Kaster, C. Suetonius Tranquillus De Grammaticis et Rhetoribus, edited with a translation, introduction, and commentary, Oxford, 1995.

Preface

xix

' thw and Dialect Gloss in the Odussia of Livius Kearns 1990: John Michael Kearns, ‘‘Semno Andronicus,’’American Journal of Philology , 111 (1990) 40
/52. Lo¨fstedt 1956: Einar Lo¨fstedt, Syntactica; Studien und Beitra¨ge zur historischen Syntax des Lateins , second edition, Vol. 2, Lund, 1956. Mariotti 1952: Scevola Mariotti, Livio Andronico e la traduzione artistica. Saggio critico ed edizione dei frammenti , 1952, second edition 1986. Marconi 1966: G. Marconi, La cronologia di Livio Andronico . Accademia Nazionale dei Lincei, 1966, (Memorie: Cl. di scienze mor. stor. e fil., Ser. 8, 12, 1966, 125
/213. Mattingly 1957: Harold B. Mattingly, ‘‘The Date of Livius Andronicus,’’ Classical Quarterly n.s. 7 (1957), 158
/163. Mattingly 1971: Harold B. Mattingly, ‘‘ Review of Marconi 1966,’’ Gnomon , 43 (1971) 680
/687. Oakley 1998: Stephen P. Oakley, A Commentary on Livy Books VI-X. Volume II. Books VIIVIII , Oxford, 1998. Palmer 1974: Robert E. A. Palmer, Roman Religion and Roman Empire. Five Essays , Philadelphia, 1974. Parsons 1999: Jed Parsons, ‘‘A New Approach to the Saturnian Verse and its Relation to Latin Prosody,’’ Transactions of the American Philological Society , 129 (1999) 117
/ 137. Pingree 1977: David Pingree, ‘‘Antiochus and Rhetorius,’’ Classical Philology , 72 (1977) 203
/223. Poccetti 1983: Paolo Poccetti, ‘‘Eine Spur des saturnischen Verses im Oskischen,’’ Glotta, 61 (1983) 207
/217. Rix 1989: Helmut Rix, ‘‘Dicthersprachliche Tradition aus vorliterarischer Zeit?’’ in Studien zur vorliterarische Periode im fru¨hen Rom , ed. by Gregor Vogt-Spira, Tu¨bingen, 1989. Sheets 1981: George A. Sheets, ‘‘The Dialect Gloss, Hellenistic Poetics and Livius Andronicus,’’ American Journal of Philology , 102 (1981) 58
/78. Suerbaum 1992: Werner Suerbaum, ‘‘Zum Umfang der Bu¨cher in der archaischen lateinischen Dichtung: Naevius, Ennius, Lukrez und Livius Andronicus auf Papyrus-Rollen,’’ Zeitschrift fu¨r Papyrologie und Epigraphik , (1992) 92, 53
/173. Suerbaum 2002: Die Archaische Literatur von den Anfa¨ngen bis Sullas Tod , edited by Werner Suerbaum, Handbuch der lateinischen Literatur der Antike , Vol. 1, Munich, 2002.

Acknowledgments

I would like to thank Professor Michael Weiss and Professor Brent Vine for their very generous comments and advice on various portions of this dissertation. The suggestions and questions of Professor Judith Ginsburg and Dr. Melanie Stowell helped, I hope, to improve its accessibility. Professor Andrew Dyck provided comments which aided immeasurably in the revision of this work for publication. To Professor Jay Jasanoff and, above all, Professor Alan Nussbaum, I owe a debt that cannot be calculated, only gratefully acknowledged. Any remaining errors of fact or judgment are, of course, my own fault.

xxi

Introduction

Although several editions of some or all of the fragments of Livius Andronicus have appeared within the last century, they contain little commentary. Also, virtually all recent major work on Livius, such as that of Mariotti, has focused on assessing his artistic talents and his ability as a translator of Homer. But, as the oldest literary Latin preserved in any quantity, the language of Livius shows many features of linguistic interest, as well, and raises many questions of phonology and morphology and a few of syntax. Therefore, I have undertaken to write a linguistic commentary on the text of Livius Andronicus. This work is not, nor is it meant to be, exhaustive. For example, some features of archaic Latin that are linguistically significant, but perfectly well treated in the standard handbooks, are dealt with cursorily, if at all, here. Finally, although Livius provides the starting-point of each comment (organized by fragment, cited according to the text of Warmington), his language does not appear to differ significantly from that of his nearest successors, such as Naevius or Ennius, and hence many of the discussions do not also end with Livius, but explore widerranging problems of Latin and, to some degree, of Indo-European linguistics.

1

Section One: The Odyssey

2 f ilie 5 Laertie ¯

Od . 2 W: ‘Pater noster, Saturni filie, Od . 5 W: ‘. . . Neque enim te oblitus sum Laertie noster , Livius preserves these examples of ii' o -stem vocative singulars in -ie% (-ii'e% ), for classical -i.1 One would certainly expect the voc. sg. of a ii'o -stem to have been -ii'e% (cf., e.g., Gk. ˆlbie2), parallel to the -e% of the o -stems. Thus -i would appear to be some sort of treatment of earlier -ii'e% . There is some evidence to show that -i- was the regular development of -ii'e% (-) in open syllables in Latin. Several places in the paradigm of denominatives in -i' e=ofrom i -stems (such as f iniire ‘to delimit’ from f inis ‘limit’) would have had this sequence. For example, the second and third singular of finiire would go back to forms of the shape *-ii' esi and *-ii' eti , which gave f inis and, prior to the regular shortening of vowels before a final -t , f init . If we compare the present of the fourth conjugation in Oscan and Umbrian, we also find - i(-) from -ii'e% (-) in open syllables–for example Umbr. 2 sg. heris, 3 sg. heri and Osc. 3 sg. sakruvit. Thus, it can be concluded that this treatment was of Italic date. But if -ii'e% (-) gave - i(-) already in proto-Italic, Old Latin cannot still have vocatives of the type f ilie . These forms cannot, therefore, be genuine archaisms, but rather have been analogically restored on the model of the o -stems (nom. sg. -o-s : voc. sg. -e% : : nom. sg. -ii' o-s- : voc. sg. -X /-ii'e% ). Umbrian has reintroduced the ‘‘regular’’ ending as well, to judge from such forms as arsie ‘sancte’ and Grabovie , according to the same analogical model. There is some evidence to suggest that, in Latin at least, - i- is also the % phonologically regular outcome of *-ii'V - in open syllables, when the short vowel is *- o% -, *- a% - , or *-u% -. In the case of *ii' o- , the first person plural of the fourth conjugation might show this treatment, e.g., f in imus B/*-ii' o% -. It is impossible, however, to guarantee that the stem vowel of f in imus has not simply been influenced by the second and third singular and second plural forms, in which the - i- is from *-ii'e% - . If *-ii' o% -mos had given something other than - i-mus , the paradigm could certainly have been leveled and - imus introduced on the model of 2 sg. -as , 3 sg. *-at ( /*-a% /), 2 pl. *-atis, beside 1 pl. -amus . ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

5

6

A Linguistic Commentary on Livius Andronicus

There appears to be better evidence, however, that *-ii' o% -, like *-ii'e% -, gave - i- in open syllables. Oscan and Umbrian have a class of verbal nouns with an oblique stem in - in-, e.g., Osc. dat. sg. leginei, acc. sg. leginum, tanginom , Umbr. abl. sg. natine. These would seem to correspond to the verbal abstract type of Latin capio, capionis ‘taking’. This inflection with -o- throughout cannot be inherited in this category, and the contrast between the Latin and the Oscan and Umbrian paradigms suggests that Latin has regularized an inherited Italic alternation of -io /- in- by extending the -o of the nominative throughout the paradigm.3 The Osco-Umbrian - in- could conceivably go back to a sequence of either *-ii' o% - or *-ii'e% -. If we compare other n -stems, beside the type of Latin sermo, sermonis , where the nom. sg. -o has also been generalized, we find also the type of homo, hominis . The -i%- in the medial syllable of hominis could theoretically be the regular reduction of any short vowel, but the diminutive homullus, rather than *homellus , should be from *-on-elo- . Therefore, it seems more likely that - in- is the reflex of *-ii' o% - and this development, too, can be dated to the Italic period.4 The treatment of the sequence *-ii' a% - is even more difficult to illustrate, but there seems to be one likely example. The Latin for ‘flute-player’ is apparently an agent compound t ib icen B/*t ibii' a% -kan- , which would seem to show that *-ii' a% developed regularly to - i- also. Finally, even *-ii'u% - could have given the same result. Lat. b igae ‘two animals yoked together’ is from *du' i-i'u% g- , a compound of *du' i- ‘two’ (/ Lat. bi- ) and the zero-grade of *i' eu' g- ‘join’ (cf. iugum ‘yoke’, iungo ‘join’). If this compound is of Italic date, the - i- of b igae would show that *-ii'u% - was treated in the same way as % the other sequences of *-ii'V -. If, therefore, as seems probable from the evidence above, the development of % *-ii'V - to - i- in open syllables was accomplished within Italic, the filie -type of vocative is not an archaism, but rather an innovation that did not survive into the classical language. ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

NOTES 1. Cf. O genitor noster Saturnie , maxime divum , attibuted to Ennius by Priscian (ap. G. L. , III.205.20 K). However, Ennius also has - i¯ in Tat i¯ (Ann . 109 W). ' mie (Ar. Th . 2. Plautus’ Bromie may well be simply modeled on the Greek form Bro 991). 3. The Oscan nominative of the type u´´ıttiuf ( B/*oi' tio¯/n/s ) has had the -n- of the stem readdded, plus an additional -s extended from the nominative singular of other stem types. 4. On the phonology of Italic *-ii' o- , see now P. Schrijver, MSS 51 (1990), p. 243 ff.

10 Morta Od . 10 W: ‘quando dies adveniet quem profata Morta est Gellius1 preserves this fragment as part of a discussion of the Fates (Fatis tribus , 3.16.9) and he understands the name Morta as equivalent to Moera ‘Fate’. There seems to be no reason to doubt his statement. The use of the verb profata est , with its suggestion of fatum ‘fate’, would appear to confirm such an interpretation. Furthermore, modern scholars2 usually associate the fragment with one or both of the Homeric passages: e¤w o ß te ke' n min j ' toio (Od. 2.99
/100) ˙ kaye' lªsi tanhlege' ow yana mo ˜ir’ o’ lo¢ j

and ' te ken d¢ ˙ o ßppo j j j

' toio (Od. 3.237
/8). ˙ kaye' lªsi tanhlege' ow yana mo ˜ir’o’ lo¢ j

If this fragment is translating one or the other of these passages, then it seems clear that Morta is serving in the capacity of mo ˜i ra. Broccia3, on the other hand, equates dies . . . quem profata Morta est with the ' rsimon mar and remarks that mo ' rsimow always means Homeric expression mo 4 ‘destined’ or ‘subject to destiny, fate’. Whatever the precise Homeric expression being translated, it seems clear that Livius has chosen Morta to function as the goddess of fate, the equivalent of moi˜ra, and hence Morta should mean something like ‘fate’ as well. Less clear, however, are the etymology and morphology of the name, to which we shall now turn. ' rsimow are from a root *(s)mer-, which also forms in Both mo ˜i ra and mo Greek the verbs me¤romai ‘get as one’s share’ ( B/*(s)mer-i' e=o- ) and ¶mmore ‘have ' row ‘fate’. one’s share’ and the substantives me' row, mer¤w ‘part, share, lot’ and mo The root is represented in Latin as well, in the verb mereo(r), which would appear to have essentially the same underlying meaning as me¤romai, but certainly a different morphological structure. The surface -e - of the present stem mere - could have resulted, in principle, from more than one source. Many verbs eventually have a stem in -e - as the result ¯

¯

¯

7

8

A Linguistic Commentary on Livius Adronicus

of a contraction of a suffix *- e´-i' e=o- . Verbs with this inherited suffix are usually divided, on the basis of their semantics, into two categories: the causatives, e.g., mone re ‘to cause to remember’ (cf. memin i), and the iteratives, e.g., torque re ‘to twist’.5 As these examples illustrate, both causatives and iteratives typically show o -grade of the root, and so one might object to analyzing mereo(r) as a causative/ iterative on the grounds of its root vocalism. But e -grade causatives and iteratives are not unknown in Latin. One such causative is vege re ‘to cause to be lively’. Among the iteratives, we find mede r i ‘to heal’ from *med - ‘measure’, although the o -grade of this root is well attested in Latin modus , modestus, and moderare . Moreover, this e -vocalism of the root, at least in vegeo and mereo(r), cannot be a very recent replacement of an o -grade since their to-participles are attested as vege%to- and mere%to- .6 In both cases the -e- of the root has prevented the medial -e%- from undergoing the regular reduction of internal short vowels to -i%- .7 Of course, in the case of mere%to- , vowel weakening does eventually prevail, giving the more familiar form merito- . Nevertheless, these forms in -e%to- are evidence that the e -vocalism of vegeo and mereo(r) was already established before the time of the normal vowel reductions and, therefore, must be of considerable age.8 Hence, mereo(r) is morphologically compatible with the causatives and iteratives, but its meaning must be considered. Since mereo(r) cannot be interpreted as ‘cause to apportion’, it would be necessary to classify it with the iteratives, if the -e - of its stem goes back to *- e´i' e=o-. On the other hand, the -e - of mere - may not be a contraction of *- e´i' e=o- at all. Another large component of the second conjugation are the characterized stative presents in -e -, e.g., tace re ‘to be silent’. But if the stem mere - belongs rather to this class, its root vocalism is again somewhat unexpected. Typically statives are formed with the zero-grade of the root, where phonotactically feasible, as in lube re and vide re . Latin can, however, show a syllabic treatment of a zero-grade when a root ends in a resonant. For example, *men- ‘remain’ would make a monosyllabic stem of the shape *mn-e -, but this becomes *mn n-e -, whence mane re . Therefore, ˚ we might rather expect a root *(s)mer- to form a stative stem *mare - B/*mr r-e ˚ 1/*mr-e -. But, as with the causative/iteratives, Latin does have a few statives with e -grade from roots of this shape, such as tene re ‘to hold’ and vere r i ‘to be in awe, to be afraid’. It would seem, therefore, that mereo(r) could be classified as either an iterative or a stative. But the fact that mereo(r) is essentially synonymous with Gk. me¤romai suggests that the two verbs may be part of a known pattern of Latin and Baltic presents in stative -e - that correspond to Greek and Indo-Iranian i' e=o- presents with similar meanings, cf. Lat. mane nre beside Av. (fra) manyeinte and, probably, Lat. ¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯

¯

Odyssey 10

"

Morta

9

horre re beside Skt. hr:s yati .9 The equation of mereo(r) and me¤romai is fairly ˚ compelling evidence for analyzing mereo(r) as a stative rather than an iterative. In either case, the expected participle *mr -to- , had it not been replaced ˚ by mere%to- , would have given morto- by regular phonological development. Semantically, the to- participle of mereo(r) should mean ‘received as one’s share’. The feminine, morta, of this stem could have been substantivized in the meaning ‘that which is received as one’s share’ and hence ‘one’s lot or fate’. This would provide a perfectly appropriate word to render moi˜ra and so Morta may preserve a trace of the original form of the participle of mereo(r). It is quite likely, however, that the name Morta would at the same time have conjured up for the Romans, as it does for many modern readers, the idea of death, mors (morti- ). Mariotti,10 while not ruling out an etymological connection between Morta and mors , imagines that this association allowed Livius to capture in a single ' toio. ˙ . . . yana word the Homeric expression mo ˜i r’ o’ lo¢j Although the semantic context of Morta would seem to require the root etymology discussed above, it is nevertheless true that the root *mer- of morior , mors , etc. formed a primary to -participle *mr -to- ‘dead’ ( /Skt. mr ta- , Gk. ˚ ˚ a )mbrotow), which would also have given *morto- in Latin, identical to the (¶ j corresponding form of *(s)mer- ‘to get a share’. If Latin had an inherited *morto- ‘dead’, Morta could, in theory, be a verbal abstract ‘death’ substantivized from this participle. Latin does have other examples of abstract nouns in -a- from to- participles, e.g., fossa ‘ditch’ (fodio), secta ‘course followed, path’ (sequor ; from an obsolete participle replaced by secu tus ). Although there is every possibility that *mr -to- survived into Italic, the ˚ synchronic to- participle of Lat. morior is not *mortus , but mortuus , a form whose origin has never been satisfactorily explained. A commonly held theory11 sets up for mortuus a pre-form *mr -tu' o- , which is said to be the result of the contamination of ˚ two forms that Italic could have inherited, *mr -to- and *mr -u' o- (/OIr. marb , ˚ ˚ W. marw ‘dead’). Such a hybrid form is apparently paralleled by OCS mri%tvu% B/ *mr -tu' o- . ˚ The difficulty with this theory lies in the assumption that an internal -tu' would have developed to -tuu' - in Latin. Anaptyxis would seem to be out of the question; nowhere else in the language does an anaptyctic vowel develop before a -u' -. On the contrary, there seems to be an opposite tendency to syncopate a short vowel before -u' - , at least when the vowel is preceded by a resonant, e.g. arvum B/ *ara-u' o-m .12 The regular phonological development of -tu' - is virtually impossible to determine due to the paucity of words that certainly contained this sequence. It has been suggested that -tu' - /-p- in Latin, but the only evidence for this development of non-initial -tu' - is an etymology that reconstructs aperio ‘uncover, ¯

¯

¯

¯

¯

¯

10

A Linguistic Commentary on Livius Adronicus

open’ as *at-u' er- , identical in root and preverb to Lith. a`t-veriu ‘open’. If correct, this would contradict the idea that mortuus could be phonologically regular from *mr -tu' o- , but the etymology is far from secure.13 ˚ One other place to reconstruct -tu' - is in the paradigm of the inherited word for ‘four’, whose stem would have appeared as *ku't-u' r - before consonants and *ku't-ur˚ before vowels. But Lat. quattuor shows a sequence -ttu- that does not directly continue either one of these. Quattuor may very well represent the result of combining the varying stems into a single, indeclinable form. The stem *ku't-urcould account for the syllabic -u- 14 and so it is tempting to think that the -tt- is the actual outcome of the -tu' - of *ku't-u' r -. If it does not somehow continue -tu' - , the ˚ -tt- of quattuor is extremely difficult to explain. Therefore, it is almost impossible to accept that -tu' - could have given -tuu' - by any regular phonological process. Since the contamination approach is plagued with apparently irresolvable phonological problems, it seems worthwhile to re-analyze the morphology of mortuus . Latin has two major types of adjectives in -uus (-uu' o- ): the deverbative type, e.g., occiduus ‘sinking, setting’ (lit. ‘falling down, cf. cado) and perspicuus ‘clear, transparent’ (lit. ‘seen through’, cf. specio),15 and the denominative type, e.g., fatuus ‘feeble(-minded)’ from the o -stem adjective *fato- ‘tired’, whose existence is implied by the substantive *fatis (in adfatim ).16 The correspondence between *fato- and fatuus raises the possibility that the inherited to-participle *mr ˚ to- may have had beside it a synonymous deadjectival mortuus . How an adjective in -uu' o- would have been created beside one in -o- is yet another question; some intermediate stage would seem to be necessary. IndoEuropean had a process by which a u-stem substantive could be derived from an o stem adjective, for example from *ai' sto- ‘hot’ (/OE ast ‘drying oven’) was derived *ai'stu- heat’ (/Lat. aestus ). In fact a tu -stem verbal abstract *mr -tu - ‘death’ is ˚ found in Arm. mah (older marh).17 There are also parallels for the creation of o ' w stem adjectives from u-stem substantives. For example, in Greek we find a ' sto a‚ stu ‘town’. Therefore, it is possible that (B/*u' astu' -o´ -) ‘townsman’ derived from j from the primary to -participle *mr -to- ‘dead’ was derived a u-stem *mr -tu- , from ˚ ˚ which in turn a new o -stem adjective, *mr tu-o- , was formed. This is the form that ˚ was inherited into Italic and Slavic. But, as discussed above, OCS mri%tvu% and Lat. mortuus differ in their immediate pre-forms. The two can be reconciled by assuming a difference in the syllabification of *mr -tu-o- . OCS mri%tvu% is the result of syllabifying *mr tu-o- as ˚ ˚ *mr tu' -o- . But in Latin, the syllabicity of the -u- was preserved by the insertion of ˚ 18 the homorganic glide -u' -, giving the pre-form *mr tuu' -o- (/mortuus ). Such an ˚ adjective derived from a u-stem was apparently synonymous with the original o stem one from which it was indirectly derived and thus mortuus was able to completely replace the expected *morto- . ¯

¯

¯

Odyssey 10

"

Morta

11

Livius’ naming of the goddess of fate as Morta may have been influenced by the surface similarity, and reasonably appropriate semantics, of mortuus , etc. But the fact that Morta is translating moi˜ra suggests that it is more likely to mean ‘Fate’ than ‘Death’ and to belong etymologically to the same family as mereo(r) , me¤romai, and moi˜ra itself. NOTES 1. 3.16.11: Caesellius autem Vindex in Lectionibus suis Antiquis : ‘‘Tria ,’’ inquit , ‘‘nomina Parcarum sunt : ‘Nona ,’ ‘Decuma ,’ ‘Morta ,’’’ et versum hunc Livii , antiquissimi poetae , ponit ex ’Odusse¤a . . . Sed homo minime malus Caesellius ‘‘Mortam’’ quasi nomen accepit , cum accipere quasi Moeram deberat . 2. Mariotti (p. 95) and Verrusio (Livio Andronico e la sua traduzione dell’ Odissea omerica [Rome, 1977], p. 45 f.), among others. 3. Recerche su Livio Andronico Epico (Padua, 1974), p. 57. ' rsimow is thus synonymous with a¶ 4. mo isimow (cf. a¶j isimon j ¶imar, p 280, Y 72, X j ' rimow (U 302) under the influence of 212) and has probably been remade from mo a‚simow. 5. Although it is useful to speak of two semantic categories, the two do share this common morphology and may ultimately go back to a single class in the protolanguage. 6. Vegeto- is found throughout classical Latin and the form mereto- is attested in inscriptions as late as 130
/90 BC (CIL 1.2.1529). 7. The non-weakening of a short vowel in a medial syllable due to the influence of a like vowel in the preceding, initial syllable is commonly known as the alacer- rule. 8. Not even mere%to- can have been the original to- participle of mereo(r) , though, which should have been formed by adding the suffix *-to- directly to the zero-grade of the root, giving a pre-form *mr -to- from the root *mer- . The participles in *- e%to- of the ˚ analogical to the *to -participles of verbs of the capio¯causatives/iteratives are rather type by the proportion *kapi' e=o- : *kapto- : : *merei' e=o- : X /*mere%to- . 9. See Schmid, p. 67. 10. Marrioti, p. 37, n. 1. 11. See W-H and Sihler (§ 185.4.a), among others. Bre´al (MSL 6 [1889], p. 127) also proposes a contamination, but of *morto- directly with its opposite v i¯vo- ‘alive’. 12. For syncope before -u' - , see further the section on pullus . 13. See, for example, the discussion of aperio¯ in E-M. 14. The generalization of the u- vowel in the word for ‘four’ is paralleled by Hom. p¤surew. 15. On this type see also the Appendix. 16. The derivation of i- stem substantives from o -stem adjectives was an IE process; ' ' compare *h2 e k -ro- ( /Gk. a ¶'krow) 0/*h2 e k -ri- ( /Gk. a ¶ kriw and perhaps Skt. a´´sri- , j j unless this goes back to *h2 ok -ri- , with Gk. ˆ ¶kriw and Lat. ocris ). j 17. See Pokorny, 4. *mer- . 18. This sort of variety in syllabification is paralleled elsewhere in IE. Within Greek, the sequence *pedi-o- (an o -stem derivative of the old locative of ‘foot’) is treated as both ' w) and *pediio- ( /ped¤on). *pedi' o- ( /pezo ' $

15 nequinont 36 inserinuntur ¯

Od . 15 W: partim errant, nequinont Graecam redire; Od . 34
/36 W: Topper citi ad aedis venimus Circai; simul $ duona $ carnem portant ad navis multam ancillae; vina isdem inserinuntur. Livius preserves two members, nequ inont and inserinuntur , of an unusual set of verbal forms with a third plural present ending -nont(ur) or the regular development thereof, -nunt(ur) . The rest of the set consists of danunt ( /dant ), explenunt ( /explent ), solinunt (/solent ), fer inunt ( /feriunt ), and ob inunt , prod inunt , and red inunt ( /-eunt ). Of these, danunt is the only one attested more than once; it occurs twelve or thirteen times in Plautus, once each in Caecilius, Naevius, and Pacuvius, and in CIL 1.2.1531.1 Prod inunt and red inunt are found in Ennius.2 The other forms, with the glosses given above, are only cited by Festus,3 who also is the source of these Livian and Ennian fragments. The long - i- in the compounds of eo is therefore metrically guaranteed,4 as is the short -a% - of danunt . The quantities indicated for the other forms will be justified by an explanation of their origin that will be presented shortly. Several theories of the origin of the -nunt have been advanced, although none seems to have won anything like universal acceptance. One of the most popular ideas takes the eo compounds as the first bearers of the ending and connects the extra -nwith that of Lith. einu` ‘go’ and/or Hitt. iyannai- ‘start moving’.5 But in those verbs the -n- recurs throughout their paradigms, whereas in Latin it is confined to the third plural.6 It is quite implausible that this distribution in Latin is coincidental. Other scholars have taken danunt as the starting point for the ending, apparently on the grounds that it is the only such form attested with any frequency. Pedersen7 and Johansson8 proposed very similar scenarios, both assuming the loss of the final consonant from dant giving a *dan to which the ending -unt was added in order to give the form more of the appearance of a third plural. This loss of final -t is a common development only in vulgar Latin and the very few archaic examples (e.g., dedron ‘dederunt ’, CIL 1.2.30) appear in inscriptions that are generally agreed to show nonRoman dialect features.9 Although this hypothesis takes into account the distributional facts, it nevertheless seems improbable, since one would have to believe that of all the third plurals in -nt , only dant lost its final consonant at such an early date. ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

13

14

A Linguistic Commentary on Livius Andronicus

Pierri’s suggestion of a cross between donant and dant simply does not work.10 Clearly, neither donant nor dant can have supplied the -unt of danunt . Vine, following Walde,11 proposed that a form *dann t (
/danunt ) was ˚ introduced as a disyllabic replacement of dant under the pressure of the disyllabic first and second plural forms, damus and datis . Aside from the phonological difficulty of getting danunt from a preform *dannt , the motivation for such a ˚ replacement is questionable. As Otre˛ bski notes, Latin consistently shows the pattern of a third plural form one syllable shorter than the corresponding first and second plurals, as in -amus, -atis, -ant ; -emus , -etis , -ent ; and -imus , -itis , -unt .12 This holds true for the active infectum forms, indicative and subjunctive, of every regular conjugation, except the present active indicative of stems in *-(i)i' e=o-, that is, the types of capio and audio. In fact, there seems rather to have been pressure to enforce the -amus , -atis, -ant type of pattern, to judge from the present subjunctive paradigm of the verb ‘to be’, in which the older s imus , s itis, sient (all disyllabic) was remade to s imus , s itis , sint . Thus damus and datis are unlikely to have influenced dant to match their syllable count; on the contrary, they may have played a role in preventing danunt from gaining widespread popularity. This tendency for the third person plural to be one syllable shorter than the first and second makes a reformation of dant to *dannt seem unlikely. Furthermore, there are no parallels ˚ for even a secondary syllabic *-n - giving -un- in Latin. ˚ Sommer, also supposing danunt to be the prototype of the group, proposed that it was created analogically according to the proportion, si%tus : sinunt : : da% tus : X / danunt .13 Also, sino is not the only verb with a third plural of the surface structure X-n-unt beside a to -participle in X-to- ; there are linunt : li%tus and cernunt : certus .14 This analogy is formally unobjectionable, but leaves open the question of why any of these verbs should have had an influence on do and then only on the third plural. To answer the first of these questions, some point of contact, other than the to participle, is needed between do and one or more of the verbs sino, lino, and cerno. Such a connection can be identified for do and sino ; the two verbs overlap semantically to some extent. The range of meanings for each includes ‘grant, allow’. This is the usual meaning of sino ; an example of this usage of do can be found at Pl. Am. 11: ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

nam vos quidem id iam scitis concessum et datum mi esse ab dis aliis, nuntiis praesim et lucro– . This semantic similarity provides the required point of contact between danunt and its proposed analogical model sinunt . The final question that must be asked in connection with Sommer’s theory is, what would have been the motivation for the creation of danunt ? Dant was obviously not intolerable, but it is undoubtedly very short and a more substantial form may have been desirable. There is reason to believe that the length of a word

Odyssey 15

"

nequinont, 36 ¯

"

inserinuntur

15

was sometimes a consideration in Latin. For example, the imperative singular of scio is always sc ito, not *sc i. Moreover, the most common -nunt forms after danunt , namely the - inunt ( /-eunt ) forms, are also associated with an exceptionally short present stem. In this connection it may also be relevant to mention the fact that the simplex forms of pleo are fairly infrequent, at least compared to its compounds, and that the third plural *plent does not seem to be attested at all. The motivating factor, therefore, seems not to have been the fact that dant was one syllable shorter than damus and datis , but that dant was just one syllable in all. Furthermore, with the exception of a single inscription, none of the -nunt forms is attested outside of the more archaic poets, if the context is known at all. This would seem to suggest that metrical considerations may have played some role. All of the occurrences of danunt in Plautus are in iambic and trochaic meters, eight of the twelve15 occur at the end of the line, where such a metrical structure would be most convenient. Also, the metrical structure of nequ inont (˘ ¯ ¯) seems to have been decidedly preferable to that of neqeunt (˘ ˘ ¯) for a trisyllabic word in this position in a Saturnian line.16 The analogy si%tus : sinunt : : da% tus : X /danunt , proposed by Sommer, provides the most convincing explanation of the origin of danunt . Once danunt had been analogically created, the other -nunt forms were very probably modeled on it. Although the to-participles may have again served as the point of contact in the cases of explenunt (: expletus ), soli%nunt (: soli%tus ) and fer inunt (: fer itus ), this will not work for the eo compounds, including nequeo, or inserinuntur ; by that analogy the forms would have come out *-i%nunt (: -i%tus ) and *insernuntur (: insertus ). How, then, are we to account for these forms? Danunt , explenunt , soli%nunt , and fer inunt are undoubtedly present tense forms and, once established as variants of dant , etc., there would have been no particular reason to continue to associate them exclusively with the to- participles. The paradigms damus , datis , danunt , explemus , expletis, explenunt , and fer imus , fer itis , fer inunt could easily have prompted a different analogy, so that - inunt was created beside - imus , - itis and inseri%nuntur beside inserimus , inseritis . The analogy does not seem to have affected the paradigm of these verbs beyond the third plural, with one possible exception. Festus preserves a form solino, for which he reports the gloss ‘consulo ’.17 If this meaning is correct, solino cannot be from the same paradigm as solinunt , which Festus glosses ‘solent ’.18 If, however, the definition ‘consulo ’ is simply an error, solino may show that the analogy did begin to proceed a bit further, but apparently did not get far before these forms fell into disuse. To summarize, the third plurals in -nunt are all modeled, by one analogy or another, on the form danunt , which is itself an analogical variant of dant created by the Sommer’s proportion si%tus : sinunt : : da% tus : X /danunt . The semantic overlap between the two verbs explains why the paradigm of sino could have

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

16

A Linguistic Commentary on Livius Andronicus

influenced that of do, although Sommer does not make this explicit. This scenario describes how danunt was created, but not why. The preference of danunt for lineend position in iambic and trochaic verse suggests that its appearance was metrically conditioned; the other -nunt forms, as far as can be determined, are found almost exclusively in poetry. Another factor seems to have been that dant was a monosyllabic verbal form */something not highly desirable to judge from, for example, the use of sc ito for *sc i. On the other hand, danunt did not survive into the classical language. This was probably due to the tendency of third plural present forms to be one syllable shorter than the corresponding first and second plural forms, a pattern whose influence can be observed in the replacement of sient by sint . This pressure seems to have reinforced the pattern of damus, datis , dant , and caused the eventual disappearance of the innovated -nunt forms. ¯

¯ ¯

¯

NOTES 1. Caecil. fab . 170 W: Patiere quod dant , quando optata non danunt . Naev. B. P. 36 W: eam carnem victoribus danunt . Pac. tr. 218
/219 W: Di me etsi perdunt , tamen esse adiutam expetunt j quom prius quam intereo spatium ulciscendi danunt . Pl. Capt. 819, Curc. 126 (dant , cdd.), Merc . 226, Most . 129, 561, Pers . 256, 852, Poen . 1253, Pseud . 767, 770, Rud . 594, Truc . 181, 245. 2. Ann. 158 W: Prodinunt famuli : tum candida lumina lucent . Red i¯nunt is attributed to Ennius by Festus, but the line is not preserved (p. 287 M). 3. exple¯nunt , p. 80 M; fer i¯nunt , solinunt , p. 162 M; ob i¯nunt , p. 189 M. 4. It is clear that the Romans considered nequeo¯, whatever its ultimate history, to be a compound of eo¯, whose inflection it follows. It will be assumed to pattern with the other eo¯ compounds in the following discussion. 5. In favor of a Lithuanian connection are Krause (KZ 69, p. 163) and Otre˛ bski (Eos 33, p. 325
/331). 6. The one possible exception, solino¯, will be discussed below. 7. IF 2, p. 302 8. Akadem. afhandl. til S. Burge , p. 30 f. 9. Although CIL 1.2.30 is from Rome, dedron has the syncope (or syllabic notation) characteristic of Oscan- and Umbrian-speaking regions, cf. dedro ‘dederunt ’ (CIL 1.2.379) from Pisaurum. See Vine, p. 339. 10. Riv. Fil. 33, p. 496. 11. Vine, p. 203; Walde, WklPh. 1915, p. 793. 12. cf. also sumus , estis , sunt . 13. Kritische Erla¨uterungen , p. 192. ' w) was the original to -participle of cerno¯ ( B/*krin14. Certus (B/*kri -to -, cf. Gk. krito e=o-, cf. Gk. kr¤nv). 15. Or nine of thirteen, if one accepts the emendation of potantes danunt for potantes dant at Curc. 126. 16. A detailed discussion of the preferred scansion of trisyllabic words after the caesura Korschiana can be found in the comment on dextra¯bus below. 17. p. 351 M; the definition is attributed to Messala augur. 18. See above, note 3.

25, 33, 34 topper Od. 23
/26 W:

. . . namque nullum peius macerat humanum quamde mare saevum; vires cui sunt magnae topper confringent importunae undae. Od. 33 W: Topper facit homones ut prius fuerunt, Od. 34
/36 W: Topper citi ad aedis venimus Circai simul $ duona $ carnem portant ad navis, multam ancillae; vina isdem inserinuntur. The adverb topper is attested in context only in the fragments of the more archaic writers. Beside the three examples in Livius above, it occurs once each in Naevius: Topper capesset flammam Volcani . (B. P. 50 W), Ennius: Topper quam nemo melius scit (Inc. 437 W), Accius: Topper , ut fit, patris te eicit ira . (Tr . 380 W), Pacuvius: Topper tecum , sist potestas , faxsit ; sin mecum velit , (Inc. 29 W), the Carmen Nelei : Topper fortunae commutantur hominibus . (fr. 5 W), and Coelius Antipater: Ita uti sese quisque vobis studeat aemulari in statu fortunae rei publicae, eadem re gesta, topper nihilo minore negotio acto, gratia minor esset. (47.1 ff.).1

17

18

A Linguistic Commentary on Livius Andronicus

These fragments are all preserved in Festus (p. 352 M) as illustrations of the definitions of topper given by Artorius (cito, fortasse , celeriter , temere ‘quickly, perhaps, swiftly, rashly’) and Sinnius (Topper fortasse valet in Enni et Pacui scriptis . . . at in antiquissimis scriptis celeriter ac mature . ‘Topper means ‘‘perhaps’’ in the writings of Ennius and Pacuvius, but in the most ancient writings ‘‘swiftly’’ and ‘‘seasonably’’ ’). The passages of Livius, Naevius, and the Carmen Nelei are quoted to exemplify topper in the sense of ‘quickly’, whereas the others are said to show topper meaning ‘perhaps’. The word is clearly archaic and seems to have fallen into disuse by the Classical period,2 so that even its meaning eventually became unclear. Neither of the basic meanings ‘quickly’ or ‘perhaps’ suits all the instances of topper . Also, it seems possible that the grammarians tried to infer the sense of the obsolete word from context. The meaning ‘quickly’ may well have been suggested by Livius’ ‘‘Topper citi . . . ’’. But the very fact that topper is followed immediately by citi argues against the two words being synonymous. ‘Perhaps’ may simply have been vague enough to cover any case where ‘quickly’ was definitely inappropriate. If, on the other hand, the definitions recorded in Festus are essentially correct, it is not very easy to see how one could be a semantic development of the other, or what earlier meaning could have undergone such a semantic split as to yield both. Since we should not, therefore, have complete faith in the statements of the grammarians, it may be profitable to use other methods to determine independently, if possible, the sense of topper . One obvious way to begin is to examine the context of the word, but this method cannot be expected to reveal much more to modern scholars than it did to the ancients. The context fails not only to confirm the translations quoted in Festus, but also to suggest any reasonably sure alternative. It may prove useful to note, however, that in seven of the nine passages, topper is the initial word of the sentence. Although word order is an admittedly unstable basis for argument, this decided preference for sentence-initial position seems, nevertheless, to be most consistent with a conjunction or adverb indicating how the sentence introduced by topper is connected to what precedes it. Unfortunately, the fragmentary nature of the evidence makes it very difficult to tell what sort of connection might be indicated. If context alone cannot provide a meaning for topper , perhaps its etymology may offer a clue. Simply by inspection, topper is probably to be classed with a group of Latin adverbs made with a suffix -per . The other members are: nu¯per ‘newly, recently, just now’ (Pl./) semper ‘always, ever, regularly’ (Pl./) parumper ‘for/in a short time’ (Enn./) paul(l)umper ‘for a short time’ (CIL 8.9642) paul(l)isper ‘for a short time’ (Pl./)

Odyssey 25, 33, 34

"

topper

19

pauxillisper ‘by slow degrees, bit by bit’ (Pl./) aliquantisper ‘for some time’ (Pl./) quantisper ‘for how long a time?’ (Caecil./) tantisper ‘for such time’; ‘for the meantime’ (Pl./) Of these parumper and paul(l)umper can be understood synchronically as parum and paul(l)um , both meaning ‘(a) little’, plus -per . In this formation, parum and paul(l)um are almost certainly the neuter accusative singulars of the adjectives,3 the accusative being the case used to indicate duration of time. Semper also contains a neuter accusative form, that of the inherited word for ‘one’, cf. Gk. ßnB/*sem . But unlike parum and paul(l)um , *sem has not been an independent element in Latin for some time and so semper must be of considerable age. Furthermore, semper certainly appears to be of the same formation as Umbr. triiu-per ‘thrice’ and Osc. pertiro-pert ‘four times’, in which a suffix -per(t) is joined postpositively to the neuter accusative of the numerals.4 Therefore, at an earlier time, semper should have meant ‘once’. The transition from ‘once’ to ‘always, forever’ is of some interest. One could imagine ‘forever’ as a single period of time extended indefinitely, but it seems impossible to trace the route of the development within attested Latin. Such a semantic development is not improbable, however, since a parallel one seems to have taken place in Germanic, where Go. sim(b)le means ‘once, formerly’, but OHG simble(s) , simblom and OE sim(b)le(s) mean ‘always’.5 These adverbs undoubtedly contain this *sem plus some suffixal material reminiscent of Lat. semel ‘once’ and simul ‘at one and the same time’.6 A mid-stage in the Germanic semantic evolution is likely to have been a sense of ‘once and for all’, which is also one of the possible translations of semel .7 The semantics of semper may well have followed a similar course. The forms in -isper , however, clearly do not contain neuter accusative forms, either singular or plural, of the adjectives. Nor could the non-neuter o- stem or a- stem accusative plurals be the basis of -isper . If the -i- is long, which seems to be the standard opinion,8 the -isper adverbs could be made from masculine or feminine i- stem substantivizations of the o- stem adjectives of the type ravis ‘hoarseness’ to ravus ‘hoarse’. While this presents no phonological difficulties, two objections can be raised. First, there is no other trace of the i- stem substantives *tantis , *paul(l)is, etc. Second, Latin already has substantives from tantus and paul(l)us , for example, namely the substantivized neuter singular forms tantum and paul(l)um. This, in turn, raises the question of what purpose a plural would have served in most of the attested Latin examples. Semantically, there is no need for one: paul(l)isper means the same as paul(l)umper , ‘for a short time’. Of the other -isper adverbs, aliquantisper , quantisper , and tantisper also refer to a duration of time. This sense, ‘for X amount of time’, would seem to have become

¯

¯

¯

¯

20

A Linguistic Commentary on Livius Andronicus

the default meaning for all relatively new -per adverbs. The one non-conformist is pauxillisper , defined by the OLD as ‘by slow degrees, bit by bit’ and attested from Plautus onward. The fact that this does not correspond to the durative sense of the productive formation implies that pauxillisper is preserving a more archaic meaning. The adjective pauxillus means ‘(very) little’ and so pauxillisper can be understood as ‘by little bits’, which would explain the apparent use of a plural form. As for the case of the form, a non-neuter accusative plural seems very unlikely. Firstly, it is hard to imagine why the neuter plural would not have been used, as in the Oscan and Umbrian parallels, if a plural form was required. Secondly, it is not clear how an accusative form, even of another gender, could be the basis of an adverb meaning ‘by little bits’. The meaning of pauxillisper is more reminiscent of the instrumental use of the ablative, as in paul(l)o ‘by a little’.9 The plural equivalent of this would give just the right sense for pauxillisper . Thus it is more likely that the - is- is the ending of the ablative plural. Nevertheless, the distinction in meaning between adverbs made from accusative plus -per and ablative plus -per seems to have become blurred fairly early in the history of Latin, since all the -isper adverbs except pauxillisper have the durative sense proper to the accusative-based type, even though they are all attested from the time of Plautus or Caecilius. Why this distinction was lost remains an open question. One -per adverb, nu per , cannot be connected with a case-form at all. Like *sem , the *nu of nu per is no longer a free-standing word in Latin, but is inherited, cf. Skt. nu and Gk. nu˜n. Sanskrit and Greek also preserve alternate forms with a ' . But it is impossible to determine whether or not Latin short vowel, nu and nu inherited both the long and short forms because the only possible traces of the short form are ambiguous. Nunc ‘now’ would scan as long, even if the inherited quantity of the *-u- were short. In any case, an inherited *-u - in *nu n-ce would almost certainly have been shortened by Osthoff ’s Law.10 Otherwise *nu only appears in the univerbation nudius which is always followed by, and often written as a single word with, tertius (quartus, etc.). Although the first syllable is marked as short in Lewis and Short, the OLD leaves it unmarked, and there does indeed seem to be no way of guaranteeing the quantity. Nudius , followed by an ordinal, occurs nine times in Plautus in iambic or trochaic verses. Therefore, it is equally possible that we are dealing with an etymological nu% dius tertius , for example, scanned nu%d? i%u s te?rtius , or an etymological nu dius tertius scanned nu¯d? i%u% s te?rtius , with iambic shortening. There seems to be no doubt that topper is also formed with the same suffix -per , although to what the suffix has been added is not immediately apparent. One theory, however, has gained extremely widespread acceptance, namely that topper is the result of an assimilation of *tod-per , in which *tod is identified with the neuter accusative singular of the pronominal stem *to- .11 Therefore, we would again be dealing with quite an old formation, since *tod , although inherited, is not attested ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Odyssey 25, 33, 34

"

topper

21

as an independent word in Latin. If this is the correct etymology of topper , one must still ask how a demonstrative stem meaning ‘this, that’ could have come to mean ‘quickly’ and/or ‘perhaps’ by the simple addition of the -per suffix. The only other trace of *tod in Latin is in the neuter pronoun istud , but other forms from the simple stem do still exist in Latin. The adverbs tum ‘at that time, then’ and tam ‘to that extent, so’ are its accusative singular masculine and feminine, respectively.12 Given that -per makes temporal adverbs, topper might be expected to mean ‘at that time, then’. This may very well have come to mean ‘just then’,13 which makes perfect sense in all the passages where the ancient grammarians translate ‘quickly’, vel sim . The sense of immediacy conveyed by ‘just then’ could easily have created the impression that topper meant ‘quickly’.14 On the other hand, the fragments of Pacuvius, Ennius, Coelius, and perhaps Accius seem to show a slightly different development of ‘then’, rather along the lines of English ‘then’ in the sense of ‘in that case, accordingly’. Thus the Pacuvius fragment */Topper tecum , sist potestas, faxsit ; sin mecum velit */could be translated, ‘‘Then let him do it with you, if he can, but if he wants to with me . . . .’’ Such an interpretation would confirm the earlier proposition that topper , with its tendency to be first in its sentence, should serve as a link to the sentence before. The connection could be strictly temporal, marking an event subsequent in time, or more consequential, indicating a circumstance that follows logically from what precedes. Above all, reconstructing a basic meaning ‘then’ for topper allows us to reconcile the apparent differences in meaning observed by the grammarians. NOTES 1. Peter, Historicorum Romanorum Fragmenta . 2. Quintilian advises against the use of topper and other archaic words: . . . nihil est odiosius adfectatione , nec utique ab ultimis et iam oblitteratis repetita temporibus , qualia sunt topper et antegerio et exanclare et prosapia et Saliorum carmina vix sacerdotibus suis satis intellecta . (1.6.40). See also Lebek, p. 36
/9. 3. Although parum exists as an indeclinable noun, historically it is the old neut. nom.acc. sg. of the adjective which became parvus , -a , -um . *Par-u' o-m became *parom by regular sound change and this gave parum , which was able to be retained as the noun/adverb because it had become dissociated from the adjectival paradigm, where the loss of *-u' - before *-o- left a stem alternating between *par- and *paru' - . The paradigm was levelled in favor of *paru' - . 4. Buck OU , §192.2. 5. Something comparable also seems to be going on in modern English, where ‘once’ and ‘ever’ can be used interchangably in some circumstances. Consider, for example, ‘‘If once we lose heart, all is lost.’’ and ‘‘In all his years at the nuclear plant, Homer didn’t once remember to wear his radiation shield.’’ In both these sorts of sentences, ‘ever’ can replace ‘once’, and yet ‘ever’ can, of course, mean ‘always’, as in ‘evergreen’. Compare also such English expressions as ‘‘once broken, never mended’’.

22

A Linguistic Commentary on Livius Andronicus 6. 7. 8. 9. 10.

11.

12. 13. 14.

Feist, p. 231. OLD mg. 3. The -i- is marked long in L-H, W-H, and Bader, for example. Paul(l)o¯ itself appears most often with expressions of time (prius , ante , post , etc.) and with comparatives. Cf. u% ndecim , for which the quantity of the initial vowel is shown by the Romance reflexes, e.g., Fr. onze . This episode of Osthoff must have been rather late, since it followed the change of *-oi' - to -u¯ - and the syncope of the internal short vowel. This idea is at least as early as Vanicˇek (1874) and is taken up by Brugmann (Die Demonstrativpronomina der Indogermanischen Sprachen [Leipzig, 1904]), W-H, E-M, L-H, etc. Also tunc ‘then’ is from *tum-ce and tandem ‘at last’ from *tam-dem . Cf. W-H, who translate topper ‘‘gerade dann ’’. This is also rather like the common translation of nu¯ per as ‘just now’. If topper had, at a sufficiently early stage, come to mean something closer to ‘quickly’, then it is not impossible that a meaning ‘perhaps’ could have developed ' xa, which is clearly from the root of taxu ' w, from this. Greek offers a parallel in ta but can mean both ‘quickly’ and ‘perhaps’. An example of the latter meaning is ' xa teu ' jeai (Op. 401). ¢ r ka¢ji tr¢jiw ta Hesiod’s d¤w m¢n gj a

30 Monetas ¯

¯

Od . 30 W: ‘nam divina Monetas filia docuit Priscian (ap. G. L. , II.198) cites this passage because Monetas shows the archaic a- stem genitive singular ending, which is fairly well-attested in Old Latin, occurring twice more in Livius (Latonas, escas ),1 twice in Naevius (fortunas, terras),2 and once in Ennius (vias),3 and is preserved in Classical Latin in the adverb alias and in the fixed, often univerbated, expressions pater (filius , etc.) familias . This seems to have been the inherited Italic ending, as it is the only one attested in Oscan and Umbrian (e.g., Osc. eituas ‘pecuniae ’, Umbr. tutas ‘c ivitatis ’). Moneta , also a cult-name for Juno,4 is Livius’ name for the mother of the ' nh ‘‘Memory’’. Mnhmosu ' nh is from Muses, his rendering of the Greek Mnhmosu the root *men( h2 )- ‘to think’ and its connection to Greek verbs from this root must have been apparent. Therefore, it is not unreasonable to suspect that a name derived from a Latin verb from this very root would have been used by Livius as the Latin ' nh. One such verb is moneo, historically a causative going equivalent of Mnhmosu e back to *mon- e´i' =o, with the characteristic root o- grade and causative suffix.5 Although one might object that moneo could no longer be synchronically recognized as akin to words having to do with remembering, there is some evidence that the Romans still believed the verb to be related to these known cognates. Varro says that meminisse is from memoria and that monere is from this same word.6 Of course, the relationship between these words is not derivational in the way Varro describes, nor is memoria even from the same root as the two verbs,7 but the fact remains that a connection is recognized. The surface similarity of moneo to Moneta caused ancient scholars to suppose that the two were related. Cicero tells the story of how Juno got the title Moneta and he uses moneo to describe her action on that occasion.8 But what process could have produced such a derivative from this verb? Latin does have a set of nouns characterized by a suffix-like element *-eto- , which has become fairly productive, usually to designate a place where a certain sort of plant-life grows, as in cupressetum ‘cypress-wood’ from cypressus . At this point it will be useful to examine the history of this nominal type in Latin, beginning with an inventory organized semantically.9 ¯ ¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Places where trees grow: aesculetum ‘a forest of aesculus ’ (Var./) arboretum ‘a plantation of trees’ (Quad./) ¯

¯

23

24

A Linguistic Commentary on Livius Andronicus

buxetum ‘a boxwood-plantation’ (Mart.) castan(i)etum ‘a chestnut-plantation’ (Col./) cornetum ‘a plantation of cornel-trees’ (Var.) coryletum ‘a hazel-thicket’ (Ov.) cupressetum ‘a cypress-wood’ (Cato/) f icetum ‘a fig-orchard’ (Var./) ilicetum ‘a grove of holm-oaks’ (Mart./) maletum ‘an orchard’ (Suet.) myrteta ‘a myrtle grove’ (Pl.)10 myrtetum (mur- ) ‘a myrtle grove’ (Pl./) nucetum ‘a nut-grove’ (Stat.) oletum ‘an olive-yard’ (Cato) ol ivetum ‘an olive-yard’ (Cato/)11 palmetum ‘a palm-grove’ (Hor./) p inetum ‘a pine-wood’ (Prop./) pometum ‘an orchard’ (Suet.) populetum ‘a plantation of poplar-trees’ (Plin.) querquetum (querc- ) ‘an oak-wood’ (Var./) rumpotinetum ‘a plantation of dwarf trees’ (Col.) salictetum ‘an osier-bed’ (Ulp.)12 ¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Places where other plants grow: dumetum ‘a clump of thorn, bushes, thicket’ (Cic/) fruticetum ‘a thicket of shrubs or bushes’ (Hor./) (h)arundinetum ‘a reed-bed’ (Cato/) iuncetum ‘a place where rushes grow, bed of rushes’ (Var.) rosetum ‘a rose-garden’ (Verg.) rubetum ‘a thicket of brambles’ (Ov./) senticetum ‘a place full of thorns or brambles, thicket’ (Pl./) sp inetum ‘a thicket of thorns’ (Verg./) v iminetum ‘a place where withies are cut, osier-bed’ (Var.) v inetum ‘a plantation of vines, vineyard’ (Var./) virgetum ‘a place full of brushwood or withies’ (Cic.) ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Places where types of stones are found: sabuletum ‘a gravel-pit’ (Plin.) saxetum ‘a stone-quarry’ (Cic./) sepulcretum ‘a graveyard’ (Catul.)13 ¯

¯

¯

Odyssey 30

"

Monetas

25

¯ ¯

Places where things other than flora or stones are found: bucetum ‘a pasture for cattle’ (Var./)14 porculetum ‘a ‘‘ridged’’ field’ (Plin.)15 ¯

¯

¯

Other sorts of places: aspretum ‘a piece of rough ground’ (Liv./) veteretum ‘a piece of ground that has stood fallow for some time’ (Col.) ¯

¯

A collection of the produce of a plant: ol iveta ‘an olive-harvest’ (Fest.) ¯

¯

Vines or parts thereof: funetum ‘a vine trained so as to form an arbor’ (Plin.)16 masculetum ‘a male or sterile growth (on a vine)’ (Plin.) ¯

¯

¯

Things that smell bad: fimetum ‘a dung-heap’ (Plin.) olenticetum ‘a foul-smelling place’ (Apul.) oletum ‘excrement’ (Veran. ap. Paul. Fest./) ¯

¯

¯

Comestible fluids: acetum ‘vinegar’ (Pl./) temetum ‘any intoxicating liquor, strong drink’ (Pl./) ¯

¯

¯

Culinary dishes:17 cocetum ‘a kind of food made from honey and poppies’ (Paul. Fest./)18 moretum ‘a dish made with cheese and pounded herbs’ (Mor./)19 tuccetum ‘some made-up savory dish’ (Pers./)20 ¯

¯

¯

An animal: rubeta ‘a kind of toad, supposedly poisonous’ (Prop./) ¯

It is immediately obvious that most of the *-eto- nouns describe land in some way, most often by what is found in it. Among these, the most common type characterizes a region by its dominant flora, and therefore this is likely to have been the area of earliest productivity.21 From the numerous ‘‘place where a certain plant grows’’ words, the use of -etum was extended to name places where other things could be found. One little set ¯

¯

26

A Linguistic Commentary on Livius Andronicus

is the rocky place words, saxetum , sabuletum , and sepulcretum . This last is a hapax , so it is difficult to determine its possible connotations, but although sepulcrum ordinarily refers to the grave or tomb itself, it can also mean the tombstone.22 Since sepulcretum would otherwise be rather isolated semantically, it seems likely that it was modeled on saxetum , which is ‘where the stones are’, to designate the place ‘where the tombstones are’. When -etum is used to name places where other things are, the words thus created mostly fall into the agricultural sphere. Cattle are to be met with in the bucetum and porcae ‘ridges of soil between furrows’ are the distinguishing characteristics of the type of field called a porculetum . Two of the -etum nouns are interesting in that they are connected only with adjectives, as opposed to other nouns, and describe areas as having the quality denoted by the adjective: an aspretum is a piece of ground that is asper and veteretum must surely be related to vetus , although the noun has a more specific agricultural meaning than merely ‘old place’. Although it is not very difficult to trace the spread of -etum from the plantplace names to the other place designations, this approach does not account for ol iveta ‘olive-harvest’, which is clearly not a place full of olives, but simply a gathering of them. Still less will it explain acetum , temetum , oletum ‘excrement’, and rubeta, which apparently have nothing to do with plants or places. These cases do not fit the productive type and therefore may be supposed to shed some light on the earlier history of *-eto- . If we examine these exceptions to the usual pattern, we notice two interesting facts. First, ol iveta and rubeta are the only non-neuter forms in the entire list.23 Not only does the appearance of both the feminine and neuter genders suggest that these nouns are really substantivized adjectives, but rubeta is actually used as an adjective modifying rana .24 Furthermore, the gender and meaning of ol iveta ‘olive-harvest’, as opposed to ol ivetum ‘olive-yard’, provide an important clue as to the original function of the substantivization. The feminine a- stems in Indo-European are widely believed to have originated as collectives derived from o- stems. It is perfectly appropriate, therefore, that a collection of olives should be an ol iveta . Second, acetum , oletum , and rubeta also stand out as the only words in the list that are not synchronically derivable from a nominal form, but rather appear beside stative verbs, namely aceo, oleo, and rubeo, respectively. Let us turn our attention now to these pairs. On any analysis, acetum , oletum , and rubeta seem to be formed from a stem in stative -e- plus a suffix *-to- . This *-to- is presumably the same suffix that makes *to- participles, but aceo, oleo, and rubeo have no synchronic *to- participles attested. In fact, few stative verbs in Latin have *to- participles at all. Even when they do exist, these participles seldom, if ever, go back directly to *-eto- . In some cases, the *-to- has been added directly to the root, as in taedeo, ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Odyssey 30

"

Monetas

27

¯ ¯

taesus (B/*tai' d-to- ). But in other cases, taceo for example, the *to- participle is tacitus , which must be from *-e% -to- (cf. Umbr. tac¸ez), not *-e-to- . The source of this *-e% -to- is almost certainly the *to- participles of the causatives/iteratives,25 since the paradigms of these and the statives fell together in the present due to regular phonological changes.26 Acetum , oletum , and rubeta would, therefore, appear to be isolated relics of a more archaic formation of *to- participles to -e- statives. If this is so, what should these forms in *-eto- have meant? If a man who is tacitus is silent and one who is cautus (B/ *kau' e%to- ) is wary, then something that is acetum is sharp or sour, which is a fitting description of vinegar. Likewise, oletum is something that is (foul-)smelling. Hence, it is very likely that the rubeta is red. The agreement of Umbr. tac¸ez with Lat. tacitus suggests that *-eto- was being replaced by *-e% -to- already in Italic, but the fact remains that, for many statives, Latin preserves no *to- participle at all. One might wonder then, what came to fill that semantic role? There seem to be two options in many cases. One is the ntparticiple; since to- participles of statives only rarely convey any sense of past time or passivity, tacitus is essentially synonymous with tacens and acetum with acens , oletum with olens ,27 etc. The other option is an adjective in -idus . These adjectives are quite productive in Latin in this function and acidus , olidus, and rubidus are attested. Temetum is most likely modeled on acetum ; both are fluids ingested by people and both are also probably grape-derived in most cases.28 Acetum may also have influenced cocetum , which Festus defines as ‘‘genus edulii ex melle et papavere factum ’’. As mentioned above, cocetum is thought to be adapted from Gk. kuke–n, which refers to a drink made from wine, as well as honey or other ingredients to taste, and so the -etum may have been added to the Greek stem when the word was borrowed into Latin, since it referred to another grape-derived beverage. If the substance later came to be thought of as a foodstuff, rather than a thick drink, or if the name cocetum was transferred to a different, more solid, concoction altogether, the formant may then have spread to the other two dishes, moretum and tuccetum . Oletum ‘excrement’ must have been the starting-point for fimetum and olenticetum . Oletum is certainly a substance spoken of as a collective, and therefore there is not much to distinguish between dung, a dung-heap, and a place that smells foul (possibly because dung is prevalent). Only a few of the attested nouns, however, can be analyzed as stative -e- plus *-to- . Most numerous are the words for clumps of trees or other plants and it is from these that -etum has been extended to certain other sorts of places, as discussed above. But the morphology of the stative derivatives of the acetum -type and of the productive place names is the same. It is very tempting, therefore, to look for a way in which the two can be connected. ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

j

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

28

A Linguistic Commentary on Livius Andronicus

There does appear to be one point of contact; *rubeto- appears in two of the semantic categories listed above, as the toad called rubeta and as the place where the rubus grows, the rubetum . While it is true that neither of these particular words is attested very early, nevertheless they should be quite old, since they show positive evidence of the adjectival origins of the *-eto- nouns and are also among the few that can be matched with a verb in stative -e- . It has already been shown that rubeta rana should literally mean ‘red toad’. By the same token, rubetum should, in the first instance, have meant ‘red thing’. Now rubus can mean both the blackberry and the prickly shrub on which it grows. Of course, despite the English name, the color of the fruit in question plausibly falls within the range of the root *h1 reu' d h-. Rubus itself goes back to *(h1 )rud h-o- and is likely to be old, since it has an exact cognate in Lith. ru`das ‘brownish-red’.29 The sequence of events can be imagined in the following way. The substantivized adjective rubetum was adopted as a word for a patch of ‘red stuff ’ consisting of one or more bramble-bushes covered with their reddish fruit. Once established in this usage, rubetum naturally became associated with rubus more than rubeo, given that Latin no longer had participles in *-e-to- as real synchronic parts of the paradigms of e- stative verbs.30 When the connection to rubeo ceased to be perceived, the -etum could not be analyzed as consisting of the stative -e- and the suffix *-to- , and the complex could then be segmented off and used freely as a suffix-like unit itself. From rubetum , the spread of -etum to other clumps of plant-life can be traced. Although the most common subset is the group derived from the names of trees, among the other plants a significant number are types of thorn-bushes: dumetum , rosetum , spinetum , and senticetum , not to mention rubetum itself. This group accounts for very nearly half of the non-tree plants and this fact would seem to lend some support to the idea that rubetum was the original model for the type. Finally, the growing popularity of -etum , in agricultural terminology especially, and particularly the influence of v inetum , seem to have prompted the creation of funetum and masculetum , which, however, do not fit the general pattern of meaning. It will have been noticed that a few words, namely olenticetum , senticetum , and bucetum , show an expanded formant -cetum . This would appear to have been extracted from cases where the sequence arose from the addition of -etum to a velar stem, as in fruticetum (from frutex ), ilicetum (ilex ), nucetum (nux ), quercetum (quercus ), f icetum (f icus ), and iuncetum (iuncus). From such cases -cetum could have been mis-segmented, although it is difficult to see why, since the velar in each case clearly belongs to the name of the plant. To return now to Moneta , some modern scholars, such as Vanicˇek, have suggested that this might be a formation like rubeta . Walde-Hofmann, on the other hand, reject the notion that Moneta can be derived from moneo at all, because the ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Odyssey 30

"

Monetas

29

¯ ¯

formation is proper only to statives, not causatives like moneo.31 But while it is undeniable that such a deverbative from a causative cannot have been inherited, it is nevertheless true that the statives and causatives in Latin have influenced each other analogically in many ways. We have already seen how participles like tacitus were modeled on those of the causative/iteratives. The analogy can also work in the opposite direction. For example, verbs of the type commonefacio and condocefacio are made from the causatives moneo and doceo on the model of another formation proper to the statives, cf. calefacio and caleo, arefacio and areo.32 Therefore, it is perfectly possible that Moneta may have been created analogically from moneo on the model of rubeta : rubeo, etc. Finally, granted that such a creation is analogically possible, what would it mean? If Moneta stands in the same relation to moneo as acetum , rubeta , and the others do to aceo, rubeo, etc., then it should be functionally equivalent to a present participle. Therefore Moneta should be the one who reminds, i.e., the memory. Recall the passage of Varro: is qui monet , proinde sit ac memoria; the semantic fit is unexceptionable. The analogy thus provides a reasonable Latin name for ' nh. Mnhmosu ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

NOTES 1. 2. 3. 4. 5. 6. 7. 8.

9. 10.

11. 12.

13. 14.

Od. 27, 14 W. B. P. 54, 46 W. Ann. 428 W. See Cicero, Div. 1.101. Other causatives of this structure in Latin include doceo¯ and noceo¯. L. L. 49: Meminisse a memoria , cum B/in/ id quod remansit in mente rursus movetur; . . . Ab eodem monere , quod is qui monet , proinde sit ac memoria . Memoria is rather from the root *(s)mer- , cf. Skt. smarate ‘remembers’. See Div. 2.69 (Quod item dici de Moneta potest ; a qua praeterquam de sue plena quid umquam moniti sumus? ), refering back to the story told in 1.101. Compare also Isidore of Seville, 16.18.8. The definitions are taken from the OLD , except where otherwise noted. This would be a feminine variant of myrte¯tum and occurs only at Vid . 93 (nescio qui servos e myrteta prosilit ). The feminine may be genuinely old (For more on nouns in -e¯ta , see below.) or may have arisen as a misinterpretation of the neuter nom. and acc. pl. The OLD suggests that Priscian has simply misquoted the text. It is somewhat interesting to note that each of the doublets olea and ol i¯va , both meaning ‘olive-tree’ and ‘olive’ has its own -e¯tum noun. This must be a place where the salix grows, and so the extra -t- before -e¯tum is presumably the result of a contamination between the expected *salice¯tum and the synonymous salictum . On the placement of sepulcre¯tum in this group, see the discussion of the semantic categories below. This word has a by-form, bu¯c i¯tum , and therefore may not really belong here.

30

A Linguistic Commentary on Livius Andronicus 15. It may be interesting to note that this is apparently derived from a diminutive, not independently attested, of porca ‘balk’, which happens to end up looking identical to the word for ‘female pig’. 16. Definition of Lewis and Short. The OLD does not give one, simply refering to the sole quotation. 17. For the most part, these words will not be taken into account in the following discussion. As will be seen from the note on each, their etymologies are not very clear and at least two of the three appear to be borrowings. 18. Definition of Lewis and Short. The OLD cites the definition from Paul. Fest. : genus edulii ex melle et papavere factum . This word is said to be, at least in its stem, a ' v ‘stir, mix’ (LSJ) and kuke–n ‘‘‘potion, borrowing from Greek related to kuka posset’ . . . containing barley-groats, grated cheese, and Pramnian wine (Il .), also honey and magical drugs (Od .)’’ (LSJ). The stem-vowel was changed to -o- in Latin under the influence of coquo¯, according to W-H; O. Keller, Lateinische Volksetymologie und Verwandtes (Leipzig, 1891), p. 81. 19. The etymology of this word is unclear. Most often it is said to be from a root *mer- , meaning something like ‘to rub’, which is supposed to be the root of Lat. morta¯rium and, with a ‘‘d -extension’’, mordeo¯. 20. According to a scholiast on Pers. 2.42, this dish is made from marinated beef or pork. It is said to be a Gaulish loan-word and possibly related to an Umbrian word toco that may mean ‘salted’. 21. These words have almost the feel of technical terms; many appear for the first time in the works of Cato, Varro, Columella, and Pliny. 22. E.g. Cic. Sen . 7.21 . . . nec sepulcra legens vereor . . . 23. I except the possibly erroneous myrte¯ta , on which see above. 24. E.g. Prop. 3.6.27. 25. The *to- participles of the causatives/iteratives are themselves analogical to verbs of the capio¯ type. The proportion was *kap-i' e=o- : *kap-to- : : *mon-e-i' e=o : X / *mon-e-to- . 26. Jasanoff, p. 66. 27. See OLD olens , meaning b, ‘rank, stinking’. 28. The etymology of te¯me¯tum is otherwise obscure; its only clear relative is abste¯mius . 29. Cf. also OE rudu ‘red color’B/*( h1 )rud h-a¯. 30. Notice that Pliny believes that the rube¯ta toad also takes its name from rubus , rather than rubeo¯ (Nat. 32.50: sunt quae in vepribus tantum vivunt , ob id rubetarum nomine . . . ). 31. W-H also object on the grounds that words of the ace¯tum -type are all neuter, with the exceptions of rube¯ta and vale¯tu¯do¯. Note that vale¯tu¯do¯ does not necessarily presuppose a lost *vale¯ta beside vale¯re , vale¯sco¯, but may be analogical to other cases where a noun in -e¯tu¯do¯ is found beside a verb in -e¯re or -e¯scere . There are a few such cases: ale¯tu¯do¯ : ale¯sco¯, inquie¯tu¯do¯ : quie¯sco¯, as- , con- , de¯- , and mansue¯tu¯do¯ : as- , con- , de¯- , and mansue¯sco¯. Nevertheless, rube¯ta certainly exists. 32. Jasanoff, p. 121. j

33 homo¯ne¯s Od . 33 W: Topper facit homones ut prius fuerunt , Warmington here prints the emendation homones , for homines , suggested by C. O. Mu¨ller. This emendation has been accepted by few, if any, other editors; Morel, Mariotti, and Verrusio, for example, all keep homines in the text.1 In addition to the familiar stem homi%n- (B/*homo%n- 2), Ennius does preserve an accusative homonem (Ann. 141 W).3 Since both homon- and *homo%n- are attested in old Latin and since Latin has both patterns, those with a short stem vowel (e.g., virgo, virgi%nis ) and those with a long (e.g., sermo, sermonis), it is difficult to say which may have been the original inflection in this particular word. Osc. humuns ‘homines ’ (nom. pl.) and Umbr. homonus ‘hominibus ’ (dat.-abl. pl.) are standardly said to be stems in -on-.4 Yet in both cases, the vowel of the suffix is spelled with precisely the same character as the vowel in the first syllable, which is unequivocally a short -o%- . It is perfectly possible, then, that Osc. humuns and Umbr. homonus both go back to a stem *homo%n-. Furthemore, in Oscan and, less regularly, in Umbrian, a long -o- tends to be spelled -u- in the Latin alphabet.5 Finally, the fact that the -o- has not been syncopated is no indication that it is long. There are several parallels to suggest that short vowels in medial open syllables can be maintained or restored under the influence of unsyncopated forms within the same paradigm. For example, the Oscan nominative singular meddı´ss has had this effect on the genitive medı´keı´s and dative medı´keı´. Such retention or restoration is quite regular in the participles in *-e% -to- , e.g., Umbr. tasetur ‘tacit i ’ and maletu ‘molitum ’.6 The only other form that is now cited as evidence that Italic had a -othroughout the stem of the word for ‘man’ is the South Picene dative singular nemu´nei´ (TE 5), which Marinetti and others have connected with Lat. ne min i.7 Nemu´nei´, written with the modified u´-sign which regularly indicates a -o- in inscriptions where the alphabet distinguishes,8 is said to preserve the older inflection in -on- , along with Ennius’ homonem , versus the -o%n- of ne min i, which is said to be a Latin innovation. While nemu´nei´ certainly seems to have a long -o- in its stem, the form is not reliable evidence for the Italic inflection of ‘man’. First, the context of the inscription does not make it at all certain that nemu´nei´ is to be translated ‘ne min i ’. It is used in apposition to the dative singular me´fistru´i´9 and may just as well be, as Eichner takes it, a proper name.10 Furthermore, if nemu´nei´ is from *ne% /hemon-, as Marinetti suggests, one might reasonably expect a long -e - contraction product to ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

31

32

A Linguistic Commentary on Livius Adronicus

be written with the modified -i´- sign, rather than the e- sign normally used for a short -e% -. Since nemu´nei´ cannot, with any certainty, be connected to hemo/homo at all, it is irrelevant to this discussion. Nevertheless, Vine, taking the Oscan, Umbrian, and South Picene forms as evidence of an Italic *homon- , has suggested that Latin inherited homon- , but replaced it with homi%n- under the influence of virgi%n-, as the only other plain n-stem for an animate being.11 Although the Oscan and Umbrian are not proof of an Italic *homon- , it cannot be ruled out as a starting point. But the pressure of the single word, virgi%n-, and a feminine one at that, is not a very satisfactory motivating force. If, on the other hand, we assume for the moment that homi%n- continues the original inflection, we must ask, what could have induced a shift to -on- ? There can be little doubt that -on- is a productive stem type in Latin, for example, in the large class of abstract nouns in -tio, -tionis . More relevant for these purposes, however, is the fact that -on- seems to have been quite productive in the creation of words for people of various occupations or characters, e.g., caupon- ‘inn-keeper’, le non‘procurer’, mu lion- ‘muleteer’, equison- ‘groom’, agason- ‘groom, driver’, erron‘truant’, etc. Therefore, there would appear to have been considerable pressure in the direction of -on- , which could have introduced homon- as a by-form of an older homo%n- . If homon- had been the original form, one might rather have expected this semantic class of persons in -on- to have reinforced the old inflection. Yet homondoes not, in fact, survive into classical Latin. Thus, if virgi%n- had any influence at all, it seems that, rather than cause homi%n- to oust an older homon- , virgi%n- may have helped to preserve homi%n- against the weight of the increasing mass of substantives in -on- . In any case, homon- would appear to be a trace of an innovation begun in old Latin that simply did not catch on. In addition to these two inflections of homo, an accusative singular hemonem is cited in Paul. Fest. (p. 100 M). Furthermore, ne mo is standardly taken to be the result of a contraction of ne% /hemo, which would provide indirect evidence for the by-form. Whether or not this is the correct interpretation of ne mo will be discussed below, but, given that Festus does preserve the one form, there can be little doubt that hemo existed. The question is, what is the relationship between hemo and homo ? Toward this end, it will be useful first to examine the etymology and morphology of homo. ' Homo is derived from the IE word for ‘earth’ *d he´g h-om (Gk. xy–n, h 'h Hitt. tekan ), which made a locatival form *d g -m-e´n ‘on earth’ (Skt. jma´n). It ' has been convincingly shown that a derivative, *d hg h-m- o˘¯n- ‘(living) on earth, ' earthling’, made from this locative is the pre-form of Lith. ˇzmuo˜ (B/*d hg h-m-o) and ' Go., OE guma, etc. (B/*d hg h-m m-o).12 ˚ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

j

¯

¯

Odyssey 33

"

homones ¯

33

¯

Lat. hemo has been explained as an e- grade variant of this derivative and is often believed to be a genuine archaism,13 while homo has long been taken as a phonological development of hemo according to a putative Latin sound change of - e´˘-to-o˘´ - before an -o- in the following syllable.14 Recently, however, the environment for such a sound change has been demonstrated by Schrijver to be considerably narrower than this one given in the older literature.15 A more precise formulation would be, -e% - became -o%- following a labial consonant and preceding a nasal consonant that is followed by another -o- .16 Therefore, since h- is clearly not a labial consonant, this rule cannot apply. In short, there is no way for the -o%- of homo to be a phonological development of -e% - in this position. Furthermore, it cannot be demonstrated that the -e- of hemo is, in fact, short at all; the only direct attestation is that of Festus, and ne mo could, of course, equally well reflect ne% /he mo as ne% /he%mo. If, however, hemo does have a short -e% -, it could be quite an old form derived from yet another locative of the word for ‘earth’. In addition to jma´n , Sanskrit also ' has a locative singular ks: a´m-i B/*d hg h-e´m-i , whose e- grade suffix might ultimately 17 account for the vocalism of hemo. On the whole, though, it would be preferable if the vocalism of the Latin, Oscan, and Umbrian forms could be reconciled with the zero-grade of their Germanic and Baltic cognates. Vine has suggested that the Italic forms do, in fact, go back to a zero-grade pre-form and that Latin (or Italic) hom- is the regular ' outcome of *d hg h -m (m)- .18 ˚ There is little to contradict the idea that *-m (m)- gave -o%m- . The immediate ˚ pre-form of the Latin superlative suffix -i(s)simo- is sometimes given as *-isa% mofrom *-ism mo- ,19 but the only evidence for an *-a% - stage is in Celtic, e.g., MW ˚ hynhaf (lit. ‘oldest’, / OIr. sinem ) B/*senisa% mo- . As Vine correctly points out, Lat. -i(s)simo- can perfectly well reflect *-iso%mo- from *-ism mo- . ˚ With this we may compare the explanation of Lat. sum and Osc. su´m ‘I am’ 20 offered by B. Joseph and R. Wallace. According to their theory, the inherited *esmi (B/*h1 es-mi ) was apocopated in proto-Italic and the resulting *esm developed ˚ first to *es m , and thence to *esom .21 Sum and su´m would both result from the generalization of the enclitic by-form *s m (and compare enclitic third singular st beside orthotone est in Lat. vocitatust (CIL 1.199.17) and Osc. destrst ‘dextra est ’.22 If such a secondary syllabic *-m - gave *-om- already in Italic, as ˚ Joseph and Wallace suggest, *-m (m)- is very likely to have had the same outcome. ˚ One other possible example of this change of *-m - to -om- is Umbr. sumel, ˚ which is synonymous, and presumably cognate, with Lat. simul . Simul 23 ‘at the same time’ would appear to be from the paradigm of the adjective similis (B/*semili- ).24 The old neuter nominative-accusative singular of similis would have been *sem-(i)li , which underwent apocope to give *seml . A secondary syllabic -l - of ˚ ˚ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

e

e

¯

¯

34

A Linguistic Commentary on Livius Adronicus

this sort regularly develops to *- l and ultimately -ul in Latin, cf. facul ‘easily’ (B/ *fak l B/*fakli ), the old neuter of facilis. Umbr. sumel seems very likely to be a neuter nominative-accusative singular form of the adjective *sem-(i)li- also, but it cannot go back directly to *semli- with a root e- grade. The surface o- vocalism of sumel can best be accounted for by reconstructing a zero-grade variant *sm l(i) of ˚ the pre-form that gives Lat. simul .25 This development of *sm - to som- in Umbr. ˚ sumel would be further evidence that *-m - became -om- already in Italic. ˚ If then hemo does not actually go back to anything inherited, and there is nothing comparable to it elsewhere in Italic, how could it have been created within Latin? The only likely source for it would appear to be ne mo. It is conceivable that from ne mo an antonymous *e mo could have been backformed on the model of nullus ‘none’ and ullus ‘any’.26 The initial h- would, very naturally, have been added, giving he mo, for the simple reason that homo was spelled so. In neither word was the h- likely to have been pronounced anyway. If this scenario is correct, the -eof hemo must perforce be long, which, as discussed above, is perfectly possible.27 If hemo was drawn in some way from ne mo, it follows then that ne mo itself cannot be the result of contracting ne% and hemo, but could simply be the phonologically regular result of the univerbation of ne% and homo.28 There seems to be no evidence to disprove that -e% -/-o%- would regularly contract to -e -, but neither are there any positive examples of such a contraction. The paradigm of the causative verbs in *-e´i' e=o- should show the treatment of this sequence; for example, the first person plural of mone re should go back to *mon- e´i' o-mos . If mone mus were the direct phonological outcome of this pre-form, ne mo could simply be from ne% / homo. But it cannot be proven that mone mus is directly from *mon- e´i' o-mos ; it is just possible that mone mus is an analogical replacement of whatever *mon- e´i' o-mos did give, modeled on the corresponding form of the other major component of the Latin second conjugation, the presents in the familiar stative -e - . This -e - is itself not a contraction product at all, but an inherited -e -. It is conceivable that, once the paradigms of the causatives and statives began to overlap by regular sound change,29 any remaining differences were leveled out by spreading -e - throughout both paradigms, thereby collapsing them into a single conjugation. Nevertheless, there is no way to rule out -e - as the contraction product of -e% - and -o%-. Hemo can thus be explained as a by-form of homo created from its opposite, ne mo, which is itself most likely a contraction of ne% and homo. Of the two stems *homo%n- and homon- , the former must be the older. The evidence for a long -o- in Italic is entirely inconclusive; the spelling of Osc. humuns and Umbr. homonus is ambiguous at best, while the context of SPi. nemu´nei´ (plus the apparent short -e% - in the initial syllable) makes the equation with ne min i dubious. Furthermore, homoncan easily have been analogically created within Latin beside an older *homo%n- , e

e

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Odyssey 33

"

homones ¯

35

¯

whereas there is no good analogical model for the creation of a secondary *homo%nfrom an older homon- . Homon- is an example of an old Latin innovation that did not persist into the classical language, but that alone is not a compelling reason to emend the transmitted reading homines in Livius. There is every reason to believe homon- to be the later of the two inflections, and no evidence of its existence before Ennius. The text of Livius is, therefore, best left unemended. ¯

¯

¯

NOTES 1. Mariotti (p. 99) cites homines in Naevius (B.P. 18 W) as support. 2. Cf. the diminutive homullus B/*homo%n-elo- . 3. The fragment is quoted by Priscian, ap. G. L. , II.206.22 K: Vetustissimi ‘homo homonis’ declinaverunt . 4. For example, at Buck OU , § 181.b. See also Untermann, p. 329 f. 5. Buck OU , § 53, 54. 6. Buck OU , § 88.2. 7. See Marinetti (p. 118), who considers the -o¯- to have been extended from the nominative singular, and Vine (p. 245). 8. Cf., also in TE 5, the nominative plural safinu´s [-o¯s ] ‘Sabini ’. AP 2, on the other hand, uses both the o -sign and the u´-sign, but there seems to be at least one case of -u´- representing an etymological -o%- , in the accusative singular meitimu´m (Marinetti, p. 90). 9. A comparative (*mef-is-tero- ) from the root *med h- of Lat. medius , according to Marinetti (p. 121). What this would mean in this context is also unclear; Eichner seems to take it as an epithet or title, although he proposes a different etymology (Die Sprache [1988] p. 200). 10. Op. cit. , p. 199. 11. p. 245. 12. Nussbaum, p. 187 ff. 13. Vine, p. 244, with reference to Nussbaum, p. 187 ff. 14. See, for example, Sommer, § 79 and L-H, § 111. 15. Schrijver, The Reflexes of the Proto-Indo-European Laryngeals in Latin , p. 466 ff. and Vine, p. 246. 16. In some handbooks, such as L-H, examples of this sound change and of the regular change of e%l /o%l are both listed under this so-called ‘‘o- umlaut’’ rule. 17. For a detailed account, see Nussbaum, p. 187 ff. 18. p. 247 ff. 19. Sihler (p. 367), for example, sets up a stage *-isa% mo- for Italic. 20. AJP 108 (1987), p. 675
/93. 21. Joseph and Wallace suggest that this secondary *- m - developed to -o%m˚ phonologically, although they do not rule out the possibility that *es m had its ending replaced with the seconday thematic ending *-om . 22. See Buck OU , § 84. 23. Semol (CIL 1.2.1531), semul (Pl. Ba. 576, 577, 591; Ci. 770; Men. 405; Mer. 689). ' niw). 24. For the raising of e% . . . i / i% . . . i , cf. cinis B/*kenis (cf. Gk. ko e

36

A Linguistic Commentary on Livius Adronicus 25. It is phonologically possible that Umbr. sumel simply has a root o -grade like Gk. ' w ‘even, equal’ (see Meiser, p. 59), but whereas o ' w is quite possibly an ßmalo o ßmalo j j ' w ‘same’, the o -grade is otherwise unattested in Italic. inner-Greek creation from o ßmo j 26. Nullus , when refering a person, and ne¯mo¯ were so nearly synonymous that the gen. nullius and abl. nullo¯ came to replace ne¯minis and ne¯mine (both found, for example, in Plautus) in the synchronic paradigm of ne¯mo¯ in classical Latin. 27. Another possible scenario, but one that would probably yield -e% - in hemo¯, is that hemo¯ was extracted from a variant spelling nehemo¯ for ne¯mo¯ (cf. pre%hendo vs. prendo¯), which was reanalyzed as composed of the privative ne% - (as in nescius , nefas , etc.) plus a hemo¯ synonymous with homo¯. However, there is no evidence that ne¯mo¯ was ever spelled this way. 28. See also Vine, p. 244. 29. E.g., in the second and third persons singular, mone¯s ( B/*mon- e´i' e-si ) and monet ( B/*mone¯t B/*mon- e´i' e-ti ) end up looking just like rube¯s (*( h1 )rud h-e¯-s ) and rubet ( B/*rube¯t B/*( h1 )rud h-e¯-t ). j

j

j

39 gavisi

¯ ¯

¯

Od . 39 W: ‘quoniam audivi, paucis gavisi ; This fragment was preserved because it shows gav is i, a non-deponent perfect of gaudere, for classical gav isus sum . Priscian attributes gav is i to the most archaic writers,1 but this is the only example of such a perfect in attested Latin and the deponent perfect is found as early as Terence (Hau. 857). This situation of early non-deponent vs. classical deponent perfect is paralleled by audere , for which aus i is once attested (Cato, Orat. 199) for the usual ausus sum .2 Both gav is i and aus i are somewhat suspect, partly because of their very limited attestation, but more importantly because they are descriptively s- perfects. The significance of this fact depends on an analysis of the morphology of gaudere and audere , to which we shall now turn. Audere ‘to dare’ is apparently a denominative present in the familiar stative -emade from the adjective avidus ‘eager’, and has undergone syncope from *au' ide- to au' de-.3 Gaudere , as well, seems likely to be a denominative to an adjective in -idus , although *gavidus is unattested. Gaudere is standardly referred to the root *ge h2 - , from which at least two presents were made in Greek. One is Hom. ga¤v ‘rejoice’ B/*geh2 -u-i' e=o-, with a ‘‘u-extended’’ form of the root that could also underlie *geh2 -u' -id ho- /*gau' ido- . It would seem , then, that *gau' ido- is an -idus adjective associated with a lost Latin *gau' ere , cf., e.g., candidus : candere. Thus *gau' ere and ga¤v would fit a recognized pattern in which Latin (and Baltic) e-statives correspond to synonymous -i' e=o- verbs in Greek and Indo-Iranian.4 The other Greek present is ghye' v/g˚ye' v ‘rejoice’ B/*ge h2 -d h-, with an added -d hwhich one is tempted to associate also with gaudere .5 If the appearance of a -d h- in ghye' v and gaudere is not coincidental, the hypothetical *gau' ido- may not be an ordinary adjective in -idus at all; rather the -i- and the -u- of *gau' ido- may be of the ' w ‘long’ (B/*dlh1 -i-gho- ), on the one hand, type found in, for example, Gk. dolixo ˚ h and Hitt. daluka- ‘long’ ( B/*dlh1 -u-g o- ), on the other. It seems somewhat odd, ˚ however, that *gau' ido- should have both a -u' - and an -i- of this sort, rather than one or the other, as with the Greek and Hittite examples above. It is probably better, therefore, to take the *-d h- (/-d- ) of *gau' ido- as belonging to the *-i-d hosuffix after all, even at the cost of separating the *-d h- of *gaudere from the *-d h(/-t h-) of ghye' v. Whatever the ultimate history of *gau' ido- , it seems to have served, like ordinary -idus adjectives, as the basis of a denominative stative. ¯

¯

¯

¯

¯

¯ ¯

¯

¯ ¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

37

38

A Linguistic Commentary on Livius Andronicus

It is to be noted that gav isus and, presumably, gav is i have a long - i- that cannot be phonological, if these are simply from *gau' id h-to- and *gau' id h-s- . There is, however, an obvious analogical source for gav isus at least; it would appear to have been modeled on v isus at a time, before syncope, when the present *gavide re could be superficially identified with videre . Given this explanation of gav isus , one might expect the same analogy to have produced a perfect *gav i d i. The absence of *gav id i, coupled with the rarity of gav is i, makes one suspect that gav is i was in fact made from gav isus itself and, therefore, is not the more archaic form. It is somewhat curious that *avidere did not go through the same analogical process(es) and receive a to- participle *av isus and a perfect *av is i (or *au' id i). One possible explanation is that the infectum forms of *avide- had already been syncopated, thus effacing any surface resemblance to videre, even before *gavidewas fitted out with its analogical perfectum. Thus the sequence of events would seem to have been: first, the forms of *avide- were syncopated; second, the analogy videre : *gavidere : : v isus : X produced gav isus (and therefrom, at some point, gav is i); third, the forms of *gavid i- were syncopated. It is conceivable that the difference in the timing of these two episodes of syncope was conditioned by the length of the vowel in the preceding syllable. This would be comparable to the situation to be discussed in the comment on pullus ‘dark’, below. ¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

NOTES ap. G.L. , II.482.9: Vetustissimi . . . gavisi pro gavisus sum protulerunt . The deponent perfect is found at least as early as Pacuvius (tr. 345 W). The root, *h2 eu' H- , is that of Lat. ave¯re ‘to be eager’, Skt. a´vati ‘promote’. The best examples to compare are Lat. mane¯re vs. Av. (fra ) manyeinte , Lat. mere¯r i¯ vs. Gk. me¤romai, and probably Lat. horre¯re vs. Skt. h r:s yati . For correspondences ˚ W. Schmid, Studien zum between Baltic e¯- verbs and Indo-Iranian *-i' e=o-stems, see Baltischen und Indogermanischen Verbum (Weisbaden, 1963). See also my comment on Morta . ' 5. cf. Toch. ka¯tk- ‘rejoice’ B/*-d h-sk - .

1. 2. 3. 4.

40 pulla Od . 40 W: vestis pulla porpurea ampla Pullus ‘dark’ is worthy of a brief note because it is one of a very few words that ' w show the treatment of an original -lu' - in Latin. The root is that of Gk. polio ' w ‘livid, dark’, both from *-i-uo- , cf. also Myc. po-ri-wa (neut. ‘grey’ and pelio ' nom.-acc. pl.), which shows that the hiatus is the result of a lost intervocalic -u' - . But pullus cannot go back to *pol-i-u' o- as well. Latin had a prehistoric episode of syncope that eliminated a short vowel between a resonant and -u' -, when a short vowel preceded the resonant ( V˘
/Ø / V˘ L_u' V, where L stands for any resonant). A secondary -lu' - resulting from this syncope simply remained, as in solvo B/*se-luu' e=o' v), and did not assimilate to -ll- . Therefore, pullus would appear to (cf. luo, Gk. lu have the -u' o- suffix, common in adjectives of color,1 attached directly to the root. ' w, For *pol-u' o- beside *poli-u' o- , compare, for example, *h1 rud h-ro- (
/Gk. §ruyro Lat. ruber), beside *h1 rud hi-ro- (
/ Skt. rudhira- ). This development can also be observed in palleo ‘be pale’ and its relations, pallidus, pallor , etc. These are derived from a stem *pal-u' o- , which also gives Gmc. *falwa- ‘pale, yellow’ (
/OE fealo ‘reddish-yellow’, ON fo˛ lr ‘pale’) and Lith. pal˜vas ‘pale, pale yellow’. The only other good instance of -ll- from -lu' - involves still another color. A certain type of Roman shoe, distinguished by its red color, was called a mulleus . This would appear to be a derivative of a stem cognate with Lith. mulvas ‘reddish, yellowish’. After a preceding long vowel, a short vowel between a resonant and -u' - was also syncopated (V˘
/Ø / VL_u' V), but at a later date than when a short vowel preceded. For example, belva ‘beast’ m ilvus ‘kite’, and pelvis ‘basin’ still scan as three syllables (belu% u' a, m ilu% u' us , pelu% u' is ) in early verse, so that this syncope can be dated within the historical period.2 ¯

¯

¯

¯

¯

¯

¯

¯

¯

NOTES 1. Cf. fl a¯vus ‘reddish-yellow, golden’, furvus ‘black, dark’, etc. 2. Instances of the trisyllabic scansion can be found at Pl. Rud. 543 (be¯lva ), Poe. 1292 (m i¯lvos ), and Laberius 94 (Ribbeck) (pe¯lvim ). See also L-H, § 141.b.a.

39

41 mandisset Od . 41 W: ‘cum socios nostros Ciclops impius mandisset , This fragment is cited by Priscian as evidence that the perfect of mando, mandere ‘chew’ is mand i. He adds, however, that some people would say that it is mandu i and still others mandid i.1 Since mand i is the only one of the three genuinely attested, it is almost certainly the real form. No other examples of the perfect exist,2 except in Charisius gram., who also gives mand i as the form.3 The other two possibilities mentioned by Priscian are easily explained as secondary creations. As for mandid i, it is doubtless modeled on the corresponding forms of verbs like condo and vendo ; the analogy would have been vendere, etc. : vendid i, etc. : : mandere : X / mandid i. Mandu i is not likely to be the original perfect of mandere . Although it has not so clear a model as mandid i, -u i is, in some sense, a default perfect that frequently takes the place of other types. For example, convertuit (CIL 8.2532) is found for convertit and reguit (CIL 5.923) for rexit .4 Therefore, it is much simpler to account for mand i being remade into mandu i than the reverse. The etymology of mandere is a more complicated business. The root *menthgiven in Walde-Hofmann cannot be right; for one reason, the root appears to have a- vocalism.5 Since a full-grade is to be expected in a plain thematic verb like ' omai ‘chew’ mandere , the root should probably be set up as *mand h-. Gk. masa would then go back to *mn d h- i' e=o- with a zero-grade of this root. ˚ If this is corrrect, Lat. mentum ‘chin’ must be unrelated to mandere . Nothing is, in fact, lost by this, since mentum has been quite convincingly connected with another root, namely *men- ‘to project, jut’, (cf. Lat. emineo ‘project, stand out’ and mons ‘mountain’).6 Therefore the pre-form of mentum can be set up as either *mento- or *mn -to- . The latter, however, would have the advantage of also giving directly ˚ the group of Germanic cognates meaning ‘mouth’ (Go. munþs , ON muðr , OE muð, OHG mund), as well as MW mant ‘jaw, mouth’. ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

NOTES 1. ap. G. L. , II.419.12 K. 2. TLL: ‘‘alias perf . non occurit ’’. 3. p. 246 K: . . . forma est quae perfectum facit secundae personae dempta s littera , velut mando mandis mandi . . .

41

42

A Linguistic Commentary on Livius Andronicus 4. See further, L-H, § 437.I.B; Sommer, § 374. 5. This fact is pointed out by E-M. 6. E-M consider this etymology of mentum to be without doubt.

46 dextrabus ¯

Od. 46 W: deque manibus dextrabus Nonius gives this fragment as an example of the use of the alternate a-stem dative-ablative plural ending -abus .1 It is often said that this ending is most often used in place of - is in order to disambiguate the genders in pairs like f ili is and f iliabus .2 In this case, however, an adjective is inflected with -abus , so gender ambiguity cannot be at issue. What, then, would have been the function of the alternative ending in this situation? It has been suggested that the use of -abus here is due to the influence of ambabus .3 The theory would seem to be that, since hands tend to come in pairs, manibus would frequently be modified by ambabus ,4 and hence -abus could have been extended to other adjectives modifying manibus. This argument seems unconvincing for several reasons. First, nowhere else does an adjective take the ending -abus because it is describing manibus . Secondly, manibus ambabus is attested surprisingly rarely and, it seems, no earlier than Apuleius.5 Thirdly, it would be odd if dextrabus were only formed after the ‘‘dual’’ adjectives ambabus and duabus precisely in a context where the hands in question are not the two belonging to an individual; obviously no one has two right hands. If a the standard association of this fragment with Hom. Od. 24.534 (tv ˜ n d’ j ¶ ra ' ntvn §k xeirv ' xea) is correct, as seems likely, the hands are ˜ n ¶ptato teu deisa those of the men of Ithaca gathered by Eupeithes to seek vengeance for the killing of the suitors. Another motivation for the form of dextrabus should perhaps be sought. Homer naturally brings to mind metrical considerations, although the Saturnian is, of course, an unstable basis for a theory of this sort. But, whatever the regulating principle(s) of the meter may be, few would dispute the observation that virtually all Saturnian lines can be divided, at a word boundary, into two segments, the first of which contains from five to nine syllables and the second from five to eight.6 Given that the fragment is from Livius’ Odyssey , deque manibus dextrabus is generally thought to be one such segment.7 Therefore, as far as the number of syllables in a half-line is concerned, dextrabus has no advantage over dexter is; both would give eight syllables. One could add here that a third possible form, dextr is , would also give an acceptable total number of syllables. Another general tendency of the Saturnian may be observed without committing to a theory of its structure. In addition to the main break in the line, another division was first noticed by Korsch, whose original formulation of the ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

43

44

A Linguistic Commentary on Livius Andronicus

rule was dependent on his own quantitative analysis of the Saturnian. In order to avoid this bias, Cole reformulated the description of the caesura Korschiana as follows: ‘‘in any half-line that contains seven or more syllables the last three or (more rarely) the last four must be preceded by word end.’’8 Thus in the half-line deque manibus dextrabus , the caesura Korschiana falls between manibus and dextrabus . Notice that although, by this ‘‘rule’’, dexter is would still appear to be as admissible as dextrabus , the other possibility, dextr is, would not be. So far there is still no reason for preferring dextrabus over dexter is. Cole goes on to examine the metrical shapes represented in the parts of the colon before and after the caesura Korschiana . He observed that when the second part of the colon has three syllables, those three can be scanned as ^ ^ in over sixty cases, as opposed to thirteen examples of ^ ^ ^. 9 Dextrabus and dexter is would both be trisyllabic words following the caesura. Either word would seem to be allowable, but the two do differ in the quantity of the medial syllable, where the statistics show a fairly decided preference for a heavy syllable. This is precisely the position in the half-line where dextrabus provides a long vowel, whereas dexter is , on the other hand, would give the less desirable short vowel. Therefore it would appear that Livius took advantage of the alternate a-stem dative-ablative plural ending -abus , normally used to disambiguate the gender of substantives, in order to create an adjectival form that would fit a preferred metrical pattern. ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

NOTES ‘Dextrabus ’ pro dexteris . . . (493.16). See, for example, L-H, § 350. L-H, § 35 end. Presumably dua¯bus would work as well for this scenario. Manibus dua¯bus is apparently even rarer and first in Ovid. See T. Cole, ‘‘The Saturnian Verse’’ (YCS , 1969), p. 10. I have found this an extremely useful study of this problematic subject and the following discussion is based on some of Cole’s findings. 7. Cole, op. cit. , p. 17. 8. Op. cit. , p. 19. 9. Op. cit. , p. 25
/28. 1. 2. 3. 4. 5. 6.

Section Two: Tragedies

4 praeda tr. 4 W:

Nam ut Pergama accensa et praeda per participes aequiter partita est,

The root of praeda is standardly said to be that of the Latin verb pre(he)ndo ‘take hold of ’, but reconciling the noun and verb involves both phonological and morphological problems that are not adequately treated in the handbooks.1 From the time of Plautus onwards, two competing forms of the verb are attested: prehendo and prendo. At the time of the texts, the distinction is not simply one of spelling; both the monosyllabic stem prend- and the disyllabic prehend- are metrically guaranteed.2 The usual explanation of the relationship between the two forms is that prend- resulted from contraction after the disappearance of -h- from prehend- . Prehend- itself is said to be composed of the preverb *prai' - and a Latin root set up as *hend- in purely descriptive terms. This treatment of *prai' - raises certain phonological questions, but the comparative evidence for the verb presents morphological problems to be addressed first. No uncompounded form of the verb exists in Latin, but it does have cognates. ' nv ‘take in, hold, comprise, The descriptive *hend- , taken with Gk. xanda contain’, suggests a root whose consonantal frame is *ghVnd-. But whereas *hendcould be the regular development of either *ghend- or *ghnd- , the Greek present ˚ cannot go back directly to either of these pre-forms, which would give a Greek *khend- or *khad- respectively, not *khand -. ' nv has an additional nasal suffix absent from the Latin, Furthermore, xanda although we also find two nasals in the Celtic, OIr. ro-geinn ‘finds room in’ and MW genni ‘be contained, find room in’ (both from a proto-Celtic *ghnd-n- 3), and ˚ some of the Germanic cognates, Go. du-ginnan and OE on- , bi-ginnan , all 4 meaning ‘to begin’. Other Germanic forms, however, lack a nasal entirely; ON geta ‘to get, obtain; beget’, Go. bi-gitan ‘to find’, OE be-gietan ‘to receive, produce’ and for-gietan ‘to forget’ reflect a Germanic stem *geta- . To account for the cognates both with and without a nasal, Pokorny sets up an IE root with two byforms, *ghend- and *ghed- . This would be an unusual situation, however, and it would be preferable if the two could be reconciled and all the attested forms derived ultimately from a unitary root. It is evidence from within Latin itself that may help clarify this picture. Praeda ‘booty’ is standardly said to be a compound of *prai' (
/Lat. prae ) and a verbal

¯

¯

¯

47

48

A Linguistic Commentary on Livius Andronicus

noun from the root of (pre)hendo itself. The absence of any -n- from this nominal form suggests that the nasal was proper only to the verbal forms and may thus originally have been simply the familiar present-forming infix. The root could therefore be set up as *ghed- , which formed a nasal-infix present *gh-n(e)-d- .5 The stem with zero-grade of the infix, *gh-n-d-, could directly give Lat. *hend- in the ˚ present; the appearance of the -n- in the rest of the paradigm (pre(he)nd i, pre(he)nsus ) is simply analogical to the present stem.6 On the other hand, an ' nv, inherited present stem *gh-n(e)-d- cannot immediately account for Gk. xanda to which we shall now turn. ' nv and perfect ke ' xanda are made from a Although the present xanda h *k and- , with invariant a-vocalism, this is clearly not the original situation in Greek. The future xe¤somai, shows the familiar digraphic spelling -ei- of a secondary *-e: - in Attic and Ionic. Therefore, the future stem is reconstructed as *khe: s- from ' n *khend-s-, by the second compensatory lengthening.7 Just as the first -a- of xanda ' v appears to be an innovation, in light of xe¤somai, so too the perfect kexanda is ' ndei (Il . 24. 192), which has the form of a secondary given the Homeric kexo perfect, with reduplication and a characteristic o -grade of the ‘‘root’’ *khend- . The thematic aorist (§)xa˘ don would be the regular outcome of a zero-grade *khn d- . ˚ This points to a synchronic ‘‘root’’ *khend- in proto-Greek with the usual -e=oablaut. The question remains: whence came this Greek ‘‘root’’ *khend- ? As we have already seen, the evidence of Lat. praeda and Gmc. *geta- suggests the PIE root was actually *ghed- , whose nasal infix present would have been of the shape *gh-n(e)-d- . There is good reason to believe, however, that the zero-grade *ghn d- had a new full˚ grade *ghend- created beside it already in the proto-language. The Gmc. -ginnan forms probably go back to *ghend-n- , whose additional nasal suffix is reminiscent of ' nv. OIr. ro-geinn and MW genni , however, can only be reconciled by Gk. xanda reconstructing a proto-Celtic *ghnd-n- ,8 and so provide no positive evidence for the ˚ remade form of the root. But the Albanian passive g je¨ndem ‘is found’, as opposed to gjej ‘find’, seems to be from *ghend- .9 It is even possible that pre(he)ndo itself goes back to *ghend- , the reflexes of *ghend- and *ghnd- in Latin being indistinguishable. ˚ Therefore, it seems that the full-grade *gh-n(e)-d- was remodeled to *gh(e)nd- within the proto-language. In Greek, this new ‘‘root’’ *ghend- served as the basis of the thematic aorist and the perfect forms. Since the thematic aorist is normally made from the zero-grade of h ' gv : ¶fu the root (cf. le¤pv : ¶li˘pon; feu ˘ gon), *g end- had the advantage of h giving a zero-grade form, *g nd- , with a syllabic element, as opposed to the zero˚ grade *ghd- of the original root *ghed- . The perfect was remade from the new ‘‘root’’ ' ndei. as well, first still with the o -grade preserved in Hom. kexo ¯

¯

¯

¯

¯

Tragedies 4

"

praeda

49

' nv itself How, then, was the secondary a -vocalism of ke' xanda and of xanda ' nv is introduced? Let us turn first to the morphology of the present stem. xanda e one of a set of Greek verbs that have a thematic present in *-an =o- B/*-n (H)-e=o -.10 ˚ This complex, apparently a thematization of *-n(e)H- , also forms presents in Armenian in -ane- , for example lk‘ane- ‘leave’, gtane- ‘find’, harc‘ane- ‘ask’, and lizane- ‘lick’.11 These appear beside what are synchronically thematic aorists, some of which are simply continuations of inherited thematic aorists, such as 3 sg. elik‘ ide, Skt. avidat ) beside lk‘ane- and gtane- .12 (
/Gk. ¶lipe) and egit (
/Gk. e† j Others, however, go back to old imperfects which have been aspectually reclassified in Armenian. Thus the aorist eharc is cognate with the Sanskrit imperfect apr cchat , ' ˚ both going back to *e-pr k -s e=o- , and elez with the Greek imperfect ¶leixe. These ˚ pairings allow us to define a descriptive rule for Armenian: from a thematic aorist, a present stem can be made by replacing the thematic vowel with the suffix complex *-n(H) e=o-. ˚ We have already seen that the Greek thematic aorist (§)xa˘ don goes back to *ghn d- e=o- with the zero-grade of the newer ‘‘root’’ *ghend- . If Greek formed ˚ present stems from such aorists in the same way as Armenian seems to have done, then the present to *ghn d-e=o- would have been *ghnd-n(H) e=o-, eventually giving ˚ ˚ ˚ *xadanv, not the actual outcome. But if we compare the other Greek presents in -ane=o- , we find that quite a few others also show a double nasal, such as ' nv, lamba ' nv, limpa ' nv, punya ' nomai, lanya ' nv, anda ' nv, tugxa ' nv, manya ' nv. It is tempting to say that *xadanv was remade to xanda ' nv in and lagxa order to conform to this common pattern, but if these presents are, in fact, derived from their aorists, the first nasal in each of these cases is equally unexpected and must therefore be secondary. Also these forms in Greek must be of relatively recent ' nv beside the still-existing le¤pv. Some further manufacture; witness limpa elucidation of this class of presents is necessary. It can be presupposed that Greek and Armenian had inherited the present *ghn d-n(H)e=o- (for the IE status of the double-nasal present, compare OIr. ˚ ˚ ro-geinn, etc., discussed above) and the aorist *ghn d-e=o- beside pairs of the type ˚ *u' ind-e=o- : *u' id-e=o- (Skt. vindati : avidat ), as well as presents in *-n (H)e=o' ˚ ' mnvB/*k m -n (H)e= - , some of which were perhaps already paired like ka with o ˚ ˚ h thematic aorists like ¶ka˘mon. The existence of *g nd-n (H)e=o-, and anything else ˚ ˚ like it that may have been present in the language at that point, made it possible for Greek and Armenian to carry out a common innovation: the *u' ind-e=o-type of present was remade by replacing the thematic vowel with the *-n (H)e=o-complex. ˚ The two languages later diverged in their treatment of the new double-nasal presents. Greek retains the nasal of both the infix and suffix, for example, in ' nvB/*linku'-n (H)e= - . But the Armenian cognate, lk‘ane- , has only the nasal limpa o ˚ of the suffix. The infix was probably eliminated to clarify the relationship between ¯

'

50

A Linguistic Commentary on Livius Andronicus

the present and the aorist, e.g., lk‘ane- beside elik‘ . Thus the Greek and Armenian double-nasal presents seem to have been modeled on the type of *ghn d-n (H)e=o- , ˚ ˚ whose two nasals were present in the proto-language, to judge from the Celtic and Germanic cognates. Let us now turn to the further history of *ghn d-n(H) e=o- itself in Greek. The ˚ ˚ ' double-nasal pattern characteristic of the type of limpa nv beside ¶li˘pon has been introduced as described above. By regular sound change, then, the present *ghnd-n(H)e=o- became proto-Gk. *kha% d-ane=o- and the aorist *ghn d-e=o- became ˚ ˚ ˚ *kha% d-e=o- . The development of the syllabic *-n- to -a% - left only a single overt nasal ˚e h% in the present with the result that *k a d-an =o- beside the aorist *kha% d-e=o- no longer conformed to the very pattern it had helped to establish. The situation was easily remedied by analogically inserting a suitable nasal into *kha% d-ane=o- to give ' nv). Meanwhile, the perfect stem *khe-khond- remained *khand-ane=o- (
/ xanda ' ndei. But the -a% - ( B/*-n -) of into the historical period, as evidenced by Hom. kexo ˚ ' nv and (§)xa xanda ˘ don must have suggested that the verb should have a-vocalism in the perfect as well, and so it was remade into classical ke' xanda. To return to Latin, praeda is generally agreed to be a verbal abstract in -a from the root *ghed- with the preverb/preposition *prai' , the same elements as pre(he)ndo. Even in attested Latin, praeda is still found as a verbal abstract, ‘the act or practice of plundering’.13 Cicero pairs it with other abstract nouns in two passages: ¯

¯

Comites vero Antoni . . . Saxae et Cafoni tradiderunt ad facinus praedamque natis , (Phil. 11.37) . . .civitatem Syracusanam propter Heraclii hereditatem non minus esse isti amicam quam Mamertinam propter praedarum ac furtorum omnium societatem (Ver. II.4.136). This juxtaposition with facinus in the first passage and furtorum in the second confirms that praeda here is a nomen actionis . Compare also Sallust’s igni magis quam praeda ager vastabatur (Jug. 55.5). But praeda was concretized quite early in the function of a patient, ‘that which is taken’ 0/ ‘booty’.14 The pre-form of praeda is often given as *prai' -heda.15 A spelling PRAIDAD is attested early (CIL 1.49) and, although Leumann-Hofmann (§ 132) speak of the process as contraction, it appears that *prai' da resulted from the syncopation of the vowel of the root. This is apparently the same syncope that produced praebere B/ *prai' -hab-e-.16 It may also be noted that an -h- left before a consonant after % syncope seems to be simply lost, thus praeda B/*prai' da B/*prai' hda B/*prai' -hVd-a. Similar cases, although without an -h- intervening between the preverb/preposition and the syncopated vowel, are praetor (PRAITOR) B/*prai' -i-tor and praemium B/ ¯

¯

¯

¯

¯

¯

¯

¯

Tragedies 4

"

praeda

51

*prai' -em-ii' o- . If such a process is responsible for praeda , then it is impossible to determine exactly what the syncopated vowel was. Secondly, this would be a type of compound (a preverb/preposition plus a verbal noun in plain -a) almost unparalleled in Latin. At best, Latin normally forms compounds with the semantics of a patient by adding the suffix *-ii' o- to the preverb/preposition and verbal root. This is the type of praemium B/*prai' -em-ii' o(emo) and adagium B/*ad-ag-ii' o- 17 (aio). Upon examining the evidence of other older IE languages, we find that Greek and Sanskrit do have compounds of a preverb/preposition plus a verbal noun in -a that function as verbal nouns. But the scarcity of such formations in Homer, and their apparent absence from Vedic, suggests that this sort of compound is not inherited. How, then, would something like praeda have been created in Latin? One way to answer this question is to compare how Greek seems to have done the same thing. Homer seems to have no instance of such a compound with e -grade of the root,18 but we find the following examples with o -grade: ¯

¯

¯

¯

¯ ¯

¯

a'ntola¤ §paoidÆ prodokÆ proxoÆ sunoxÆ Beside §paoidÆ and proxoÆ, the simplex forms a'oidÆ and xoÆ are also attested in Homer. These latter are verbal nouns of the well-known tomÆ-type, with o -grade of the root and the suffix -a. Therefore, it seems reasonable to think that xoÆ, for example, has played some part in the creation of proxoÆ. The fact that both xe' v and proxe' v are found in Homer suggests one, although not the only, possible explanation of the proxoÆ-type. As is well-known, preverbs are not always attached to their verbs in Homer and hence univerbation is clearly not very old. But once the univerbation took place, the model of xoÆ beside xe' v could easily prompt the creation of proxoÆ beside proxe' v. So it makes sense that Vedic, which shows almost no univerbation of preverbs and verbs, also has no nominal forms like proxoÆ. Given that *prai' -heda is not an inherited type of compound, the e -grade cannot be established on that basis. Furthermore, no simplex *heda is attested and ¯

¯

¯

52

A Linguistic Commentary on Livius Andronicus

there seems to be no evidence that e -grade verbal nouns of this type existed in Latin. In short, there is no clear reason to think that the vowel in question was *-e- at all. Latin does, however, preserve several clear examples of the tomÆ-type of verbal noun, e.g., toga : tego

¯

rota : OIr. rethid ‘runs’ mola : OIr. melid ‘grinds’19 These are residual and heavily concretized, but they do show that Latin may once have had a verbal noun of this type from the root *ghed- . It is true that no independent *hoda exists, but, unlike *heda, this would at least be of a known inherited type represented in Latin. If Latin formed compounds parallel to the Greek proxoÆ-type, we might expect a pre-form *prai' -hoda. Therefore, it seems more likely that the syncopated vowel of praeda was an *-o- , rather than an *-e- . If we look for other examples of the praeda -type of compound in Latin, we find that there are very few candidates with reasonably sure etymologies. Perhaps the best example is ap(p)luda ‘chaff ’ (Naev./), which is almost certainly to be connected with the verb plaudo.20 The preverb is most likely to be at , which, like abs- , denotes origin and/or separation. This preverb/adverb is only marginally preserved in Latin. As a preverb, it appears in atavus ‘great-great-great-grandfather’21, which can be straightforwardly explained as a possessive compound meaning ‘having a grandfather (descended) from (him)’.22 In a more purely adverbial usage, at ‘but’23 (B/*ati ) can be understood as having developed, from the more separative meaning of the preverb, along the same semantic lines as sed ‘but’, which contains the preverb/preposition s e¯˘ - of secedo ‘go apart’, securus ‘free from care’, etc.24 Therefore, at- would give the right sense of separation or dispersal for ap(p)luda and would no doubt assimilate to give app- in this position.25 Remora (Pl./) ‘delay’ is descriptively a compound of this type, but cannot be very old since the internal short vowel has escaped reduction. Remora could easily have been created relatively recently by the simple analogy, moror : mora : : remoror : X
/remora . This analogy would be parallel to the one proposed creating proxoÆ beside proxe' v on the model of xoÆ beside xe' v. It seems highly likely, therefore, that *prai' -hoda was made, directly or indirectly, from a *hoda also of the tomÆ-type. To return to pre(he)ndo, one might have expected the same phonology that produced praeda B/*prai'-hoda to have operated on the verb as well. Why, then, is it pre(he)nd- , rather than *prai' nd- B/*prai' -hend- ? Prehend- is standardly said to show ¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

Tragedies 4

"

praeda

53

that the diphthong of *prai' - has become a monophthong, which was then shortened before the vowel giving pre% -; and prend- is explained as a contraction of prehend- .26 There are examples of such short scansions of -ae- in hiatus; LeumannHofmann (§ 118) list the following examples: praeut (often in Pl.) praeust is (Verg. Aen . 7.524) ¯

Pellaeo (Pl. As. 333 vs. 397) ¯

Lindsay also cites praeolat with a short -ae- (Pl. Mil. 41).27 It is notable, however, that these are sporadic instances, and thus not truly comparable to prehendo, in which the pre- never seems to be scanned as a long syllable. In addition, the occasional shortening of -ae- in praeust is, etc., has apparently had no effect on the spelling, whereas prehendo is not usually spelled prae- .28 There is some evidence to suggest that *prai' -hend- was, in fact, syncopated to *prai' nd- , as the comparison with praeda would suggest, and that pre(he)nd- is secondary. Festus (p. 166 M) gives the gloss: Nancitor in XII nactus erit , praenderit . Notice that praenderit is not the strange form that Festus is explaining; it is one of the two ways he glosses the archaic nancitor . Of course, Festus is not an archaic author himself and so one might well hesitate to attribute such antiquity to a word found only in Festus’ active vocabulary. On the other hand, he was certainly an antiquarian familiar with all sorts of linguistic archaisms and oddities, so it may well be that he picked up praendo from some considerably older source no longer extant. If Festus’ praendo preserves what might be expected, on the basis of praeda , to be the phonologically regular outcome of *prai' -hend- , pre(he)ndo still remains to be explained. In attested Latin, further combinations of pre(he)ndo with an additional preverb, such as appre(he)ndo, compre(he)ndo, depre(he)ndo, and repre(he)ndo, are very common. If we assume for the moment that *prai' -hend- had gone to *prai' nd- , what would happen when another preverb was added? In a form such as *kom-prai'nd-, the diphthong *-ai' - finds itself in an internal syllable; and there is an apparent sound law in Latin by which *-ai' - becomes *-e- in an internal syllable before a nasal consonant.29 The clearest example is perhaps obscenus B/*-skai' -no- , ' w. Without this sound law, moreover, with the root of scaevus and Gk. skai(W)o there is no obvious explanation of alienus .30 With it, however, it is possible to hypothesize that alienus is a derivative in *-no- of the locative *aliai' of the a-stem alia (cf., on the one hand, the a-stem genitive singular alias in adverbial function ' w and gen. sg. v ic i : adj. and, on the other, e.g., loc. pe' rusi : adj. perusino 31 v ic inus). ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯

¯ ¯

54

A Linguistic Commentary on Livius Andronicus

By this rule, then, *kom-prai' nd- would become *kom-prend- . From this and the other compounds, a new simplex *prend- could have been extracted, which would account, more or less directly, for the shorter form prendo. The only other intervening change would be the shortening of the *-e- of *prend- by a relatively recent application of Osthoff’s Law, which shortens long vowels before a sonorant followed by another consonant.32 Although the length of the -e- of prendo cannot, of course, be determined from the Latin, the Romance reflex (Fr. prendre ) seems to require a short -e% -. Since prendo cannot, therefore, be a contraction of prehendo, the longer form must still be accounted for. Quintilian (1.5.21) describes what he considers an archaic practice of spelling -e- as -ehe- that persisted, in some cases, even to his own day: ¯

¯

¯

¯

¯

¯

¯

¯

¯

Inde durat ad nos usque vehementer et comprehendere et mihi , nam mehe quoque pro me apud antiquos tragoediarum praecipue scriptores in veteribus libris invenimus .33 If the *-e- of *prend- remained long for any length of time before being shortened by the ‘‘neo’’-Osthoff, there could have been the opportunity for prehend- to have been introduced as a way of spelling *prend- . Thus after *prend- was shortened to prend- , the two spellings simply remained. The disyllabic scansion of prehendwould thus have to be a later spelling pronunciation, in apparently the same way that vemens and vehemens are both attested. In summary, praeda is the outcome of a syncopated *prai' -hoda, rather than *prai' -heda. This *prai' -hoda was, in all likelihood, made from a verbal abstract *hoda of the same type as Lat. toga , etc. Syncope also operated on the verbal compound, giving *prai' nd- (cf. Fest. praenderit ) from *prai'-hend- . When an additional preverb was added, *-prai' nd- became *-prend- by an apparently regular Latin sound change, by which *-ai' - becomes *-e- in an internal syllable before a nasal consonant. From such compounds as *kom-prend- , etc., a new simplex *prend- was extracted and this form, with what seems to be an archaic spelling of -e- as -ehe- , is probably preserved in prehendo. The by-form prendo, on the other hand, would seem to represent the phonologically regular outcome of *prend- after the -e- underwent an Osthoff-type shortening. ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

NOTES 1. For this root etymology, see W-H, E-M, etc. 2. Lindsay, Early Latin Verse (Oxford, 1922), p. 151. 3. Thurneysen, KZ 63 (1936) p. 115.

Tragedies 4

"

praeda

55

4. The semantics of the Germanic verbs are comparable to Lat. incipio¯, that is, ‘take on, up’ 0/ ‘begin’. 5. This is an unusual root shape to make a nasal infix present, but compare *bhegwhich makes Skt. bhanakti ‘breaks’. 6. Such analogical spreading of a present-forming nasal infix throughout a paradigm is common in Latin, cf. iungo¯, iunx i¯, iunctus (but nominal iugum ) vs. Skt. pres. yunakti , aor. ayujat , perf. yuyoja , ppl. yukta- . 7. It is very difficult to see how this isolated e- grade could be an innovation; it is much more likely to be a relic of an older state of affairs. 8. See above, note 3. 9. G. Meyer, Etymologisches Wo¨rterbuch der Albanesischen Sprache . 10. Rix, p. 211. 11. Klingenschmitt, Das Altarmenische Verbum , p. 164 ff. 12. The following scenario describing the history of the Armenian *-an-e=o- presents and ' nv in the creation of the Greek double-nasal presents was the role of xanda outlined to me by J. Jasanoff in a personal communication. 13. OLD 1c. Note, however, that in the passage of Accius (tr. 271 W: passimque praedam pecua vallebant agris ) given under this heading, praeda , especially since it appears to be in apposition to pecua , could well mean simply ‘booty’. Thus I would translate, ‘and on all sides they walled in the flocks taken as booty’. 14. OLD 1a (CIL 1.25/), from which ‘potential booty’ (1b), ‘prey’ (2), and ‘reward, prize, profit’ (3) are reasonable developments. 15. Pokorny, L-H § 132, Bader § 102, etc. 16. Praehibe¯re is almost certainly recompounded from the preverb and existing simplex habe¯re and the scansion of praehib- is ambiguous between ˘ ˘ and */. Also, praebia ‘prophylactic amulets’ may be from *prai' -hab-ii' o- , but the etymology is not entirely clear. 17. The long -a¯- here must be analogical, as it is in other members of this morphological category, such as conta¯gium , and in similar formations, such as amba¯ge¯s and compa¯ge¯s . 18. The source of the following information on Homeric nominal formations is Risch. 19. Mola is reconstructed with an o- grade for morphological reasons, although *mel- or *mol- would ultimately give mol- in Latin. Hence, molo¯ is likewise ambiguous and could go back to an e- grade (cf. OIr. melid ) or an o- grade (cf. Go. malan ). 20. E-M simply dismiss the word as non-Roman. W-H, on the other hand, connect applu¯da with the verb plaudo¯ and the preverb/preposition is generally assumed to be abs- . Although a concretization of *abs-plou' d-a¯ with a passive meaning, like praeda , would give a good description of chaff as ‘that which is beaten away’, this pre-form raises a phonological problem. The regular outcome of *abs / p- is as- , as in asporto¯ ‘carry away’ and aspello¯ ‘drive away’. A variant spelling adplu¯da is, in fact, attested in Pliny (Nat. Hist. 18.99.6) and if the preverb/preposition of ap (p )lu¯da were ad- , the phonology would work, but the semantics would be less than satisfactory. 21. W-H connect atavus with*at (i ). E-M, on the other hand, accept the etymology of Festus: at [t ]avus , quia atta est avi , id est pater . . . . But if an atavus is to be understood as the father of an avus , he would only be a great-grandfather. 22. As such, atavus is entirely parallel to, although more archaic than, abavus ‘greatgreat-grandfather’.

56

A Linguistic Commentary on Livius Andronicus 23. Also found as the first part of atque , atqu i¯. Among the cognates are the first parts a uo n’ (E-M, W-H) as well as OIr. aithof Gk. a't-ar ‘but’ and Go. a þþan ‘ a'll¢ j ‘re- , ex- ’ (W-H, Thurneysen, p. 500). 24. As a sort of postposition, *-at has contracted, apparently with the stem vowel, to give the inherited o- stem ablative singular in *-o¯d (Lat. GNAIVOD, Skt. deva´¯t , Lith. die˜vo , etc.)B/*-o-at . For a more detailed discussion, see Stang, Vergleichende Grammatik der Baltischen Sprachen . 25. Pliny’s adp- is probably secondarily extracted from app- (W-H), since most cases of surface app- were easily analyzable as *ad / p- , whereas at- was no longer a synchronically recognizable preverb/preposition. 26. L-H, § 118, 133; Lindsay, op. cit. p. 151. 27. Op. cit. , p. 151 28. The spelling praehend- is found, but apparently only consistently in Justinian and the Scriptores Historiae Augustae , where it has probably been reconstituted. 29. I am indebted to A. Nussbaum for suggesting the possibility of this sound law. 30. L-H (§ 294.2.b) follow the suggestion of Skutsch (Glotta 3, p. 355) that alie¯nus is dissimilated from *ali- i¯no- , but the only parallel given is lanie¯na (taberna ) ‘butcher’s shop’. But the etymology of this word is unclear and the putative derivational morphology, *-ii' o- (lanius ) : *-i- i¯no- ( 0/lanie¯nus ), is highly questionable in view of Latium : Lat i¯nus , etc. 31. Terre¯nus could also be derived from the locative of the a¯- stem, terra , but is usually taken to be from the neuter s- stem (OIr. ti´r ‘land’) with -e¯no- from *-es-no- (cf. ae¯nus ‘brazen’ with aes ‘bronze’). See L-H, § 294.2. 32. cf. u% ndecim (Fr. onze ) B/*u¯nidecim B/*oi' no- . 33. The inclusion of mihi (cf. Umbr. mehe ) here is misleading, since it is actually older than m i¯, which is contracted from it.

20 opes, opitula ¯

tr. 20
/22 W:

¯

Da mihi hasce opes quas peto, quas precor! Porrige, opitula!

The etymological dictionaries1 standardly connect ops with a large assortment of words, both within Latin and without, whose relationships may now be clarified on the basis of recent work in the phonology of the Anatolian languages. One meaning of ops, ‘wealth, resource’, accords well with a derivational family in Hittite, consisting of the denominative verbs h% appina h% h% - ‘enrich’ and h% appinesˇ‘become rich’ and the adjectives h% appina- , h% appinant- ‘rich’. That this family is related to ops , I do not believe has ever been denied. The same may be said of Skt. apnas- ‘wealth’ and Av. afnahvant- ‘wealthy’. Slightly removed semantically, but almost certainly to be associated with these, are Hitt. neut. nom.-acc. sg. h% appi´r , h% appar ‘business, trade; compensation, price’, h% apparai- ‘trade, sell’, h% appir(iy)a‘city, town’ (as the center of commerce; the original meaning was probably ‘marketplace’), Lycian epirijeti ‘sells’, etc. The Hittite evidence shows that the root had an initial laryngeal, either *h2 - or *h3 - (*h1 - being lost in all positions in Hittite). Therefore, Lat. op- could go back to *h2 op- , *h3 ep- , or *h3 op- . Although Hittite is of no help in determining which of the two laryngeals is correct in this position, Lyc. epirijeti with an initial vowel shows that the laryngeal in question must have been *h3 -, since an initial *h2 - remains in Lycian, e.g. xn˜ tawa- ‘rule’ with Hitt. h% ant- ‘front’ B/*h2 ent- .2 So we are left with a choice between *h3 ep- or *h3 op- for Lat. op- . Of course, *h3 ep- would have been colored to *h3 op- already in the proto-language, but the original nominative singular is attested as ops in Accius, (Inc . 5 W),3 and in the divine name Ops (Consiva) .4 Therefore we are dealing with a feminine root noun verbal abstract, originally meaning ‘abundance’, from a verbal root meaning ‘to abound’. In such a noun from a root of this shape we would expect e/z- ablaut (cf. nex ), so ops is more likely to reflect *h% op-s from, in the last analysis, *h% ep-s .5 The other cognates would then be reconstructed as follows: Skt. apnas- would go back to *h3 ep-no-s- or *h3 op-no-s- .6 The same s-stem with the IE possessive suffix *-u' ent- underlies Av. afnahvant- . Hitt. h% appina- is the reflex of this denominative adjective in *-no- and goes back to *h3 ep-en-o- , which is in turn derived from an r /n -stem substantive (cf. Skt. tapana -B/*tep-eno- ). As often in Hittite, the adjective

57

58

A Linguistic Commentary on Livius Andronicus

received an -nt -extension, giving h% appinant- . Hitt. h% appi´r is from *h3 p- e¯´r .7 Finally, h% appar(iy)a- looks to be from *h3 p-er-o- . The handbooks almost universally connect this derivational family with Lat. opus and a group of apparently cognate words with meanings related to ‘effort’ or ‘work’, particularly of a ritual nature. But the reconstruction of the root *h3 ep- for ops effectively rules out the possibility of such a connection. While opus itself phonologically could be from *h3 ep-e/os- , it has been suggested that opus should be from the same root as epula(e)/-um ‘banquet’, which seems to have originally meant ‘a religious work or performance’,8 a meaning which was later specialized to mean ‘a sacrificial meal’, then ‘(public) feast, banquet’. If, therefore, opus and epula(e)/-um are derived from the same root, this root cannot be *h3 ep- because this would have become *h3 op- already in the proto-language and epula(e)/-um , with its initial e- , could never have come from this. The root of this family would therefore have to be set up as *h1 ep- , *h1 - being the only laryngeal which would not color a following short -e% -. Epula/epulum would ultimately be an instrumental noun in -la/-lum (cf. regula , stragulum ) with anaptyxis.9 Opus itself would, in that case, go back to *h1 opwith o -grade of the root. Although a root e -grade is expected in neuter s-stems, it is a fact that Latin has several o -grade s -stems, e.g. pondus , foedus, mu nus. Also to be taken with this root are Skt. a´pas- ‘work’ B/*h1 e/op-os- , Ved. a¯´pas- ‘cult-act’ B/*h1e/op-os- and a¯´pra- ‘sacrificer’ B/*h1e/op-ro- , and OHG uoben ‘to celebrate a festival’ and uoba ‘festival’ B/*h1op- . Since the root shapes are demonstrably different, none of these words can be related to ops. Opitulo(r) 10 clearly contains the stem of ops and is most simply interpreted as a denominative from the adjective opitulus , which is attested as a cult-title of Jupiter in Paul. Fest. (p. 184 M.).11 Opitulus itself would seem to be a compound, of the same type as foedifragus , from the root *tel h2 - without, of course, the nasal infix of the present tollo B/*tl -n- h2 - . ˚ The question of whether Lat. ops and opus are etymologically related can now, perhaps, be settled, with the application of discoveries made in the area of Anatolian phonology. The semantics of ops strongly suggest a connection with Hitt. h% appina- , etc., which means that this family must be reconstructed with an initial *h2 - or *h3 - . But this semantic group also includes Lyc. epirijeti , which can only have had an initial *h3 -. Ops and its cognates, therefore, must go back to a root *h3 ep- . On the other hand, Lat. opus belongs to a family of words with semantics suggesting ‘work’, often of a religious sort. Most of these words (e.g., opus, Skt. a´pas- ) are also compatible with *h3 ep- . But since it is very likely that Lat. epula(e)/ epulum is etymologically connected with opus , this family must have had an initial *h1 -. Lat. ops and opus must, in that case, be etymologically unrelated and go back to two distinct roots *h3 ep- and *h1 ep- , respectively. ¯

¯

¯

¯ ¯

¯ ¯

¯

¯

Tragedies 20

"

opes, opitula ¯

¯

59

NOTES 1. See, for example, Pokorny or W-H, but also such recent works as Tischler’s Hethitiches etymologisches Glossar (Innsbruck, 1977
/). 2. S. Kimball, ‘‘*H3 in Anatolian’’ Festschrift Hoenigswald , p. 185
/92 (although Kimball considers both Lat. ops and opus to be from *h3 ep- ). See also now Melchert, p. 72. 3. Priscian (ap. G.L. II.321.24), cites this fragment (quorum genitor fertur esse ops gentibus ) saying that ops is an adjective used by archaic writers to mean opulentus , but here has the meaning ‘‘opem ferens et auxilium ’’. The fact that in this passage ops does not have what Priscian gives as the usual meaning seems to indicate that he has confused an example of the root noun with an adjective ops also attested in Paul. Fest . (p. 191 M: Ops antiqui dicebant opulentum , unde e contrario inops ). This adjective is probably to be explained as a back-formation from inops . 4. Varro, LL 5.57. The nom. sg. also appears as Opis (Pl. Bac. 893, Hyg. Fab. 139.1, Paul. Fest ., p. 187 M), but this has been remade in the same way as fru¯ gis from frux and vehis from *vex . 5. J. Schindler ‘‘L’apophonie des noms-racines Indo-Europe´ens’’ BSL 67, p. 37 f. 6. A neuter adjectival abstract in -s- can be derived from a thematic adjective, cf. Gk. ' w ‘disabled’. ph˜ row, -eow ‘diability’ from phro 7. Its expected locative h% appi´ri B/*h3 p- e´r/i is attested, but also a locative h% appari , which appears to be remade analogically on the basis of the stem of the other oblique forms, *h3 p-( r)r- . 8. See E-M for˚this meaning of epula(e)/-um . 9. See L-H, § 283. 10. Note that this fragment of Livius is the only place where the active inflection is attested. 11. Bader, § 97.

29 praestolaras ¯

¯

¯

tr. 29 W: Nimis pol inprudenter servus praestolaras. Nonius (p. 475 L) cites this fragment to illustrate the active inflection of the verb praestolo(r) as opposed to the deponent inflection regular in classical Latin and found as early as Plautus and Terence. But a non-deponent form is also attested in Turpilius Com. 153. This issue of deponent versus non-deponent inflection has been used as evidence in the larger and much-debated question, to which we shall now turn, of the morphology and etymology of praestolo(r) . In spite of these early active forms, Stowasser suggests that the deponent inflection is original on the basis of his analysis of the verb as a back-formation from an expression praestolatus , meaning ‘having brought oneself [so as to be] at hand (praesto)’.1 Stowasser detects an element of motion in the meaning of the adverb praesto on the evidence of a passage of Tibullus which he gives as ‘‘tibi praesto pauper adibit ’’ (1.5.61). From this he argues that the adverb could have been paired with other verbs of motion, including fero used in the (medio-)passive.2 The argument would be somewhat stronger if there were an example of the phrase (praesto ferr i) whose perfect (praesto latus ) is the alleged starting point for the back-formation. Nevertheless, if this scenario were correct, one might expect the perfect participial forms of praestolo(r) to be attested in early texts more frequently than the present stem, but this is not the case.3 There are also phonological problems with the idea of a univerbation of praesto and latus . The combination should be reasonably old, given the extreme scarcity of the participial forms to which praestolo(r) is said to have been back-formed. Its antiquity is also implied by the necessary assumption that the present praestolo(r) has been created and is so well entrenched that there is no longer any trace of an expression praesto fero. For all this to have already taken place, the univerbation should probably have occurred at a time when the forms involved were praesto (or praestod 4) and *tlatus (B/*tl h2 -to- ). If so, the initial tl- of *tlatus would have been ˚ in a word-internal position and should have given, at first, *-kl- (cf. poc(u)lum ‘drinking-vessel’ B/*poklo- B/*po-tlo- ). Of course, if the original univerbation were still transparent enough after tlatus had become latus , it would have been a simple matter to remake the univerbation on the analogy of the free-standing form. But if the component parts of the univerbation could still be identified in order to remake praestolatus (instead of *praestoc(u)latus ), they cannot at the same time have been so opaque as to generate a new present praestolo(r) . ¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

61

62

A Linguistic Commentary on Livius Andronicus

Perhaps even more damaging to this theory is the fact that the Tibullus passage does not really offer any proof of the use of praesto with a verb of motion. Stowasser’s presentation is rather misleading. The passage, according to the 1988 Teubner edition by G. Luck, reads: ¯

Pauper erit praesto semper tibi , pauper adibit primus et in tenero fixus erit latere , Although the text is problematic at this point, most editors (e.g., Luck, Putnam, Smith, Postgate) punctuate the line with either a comma or a colon before the second pauper . The balance of the phrases certainly seems to require that praesto be taken with the preceding verb.5 Furthermore, even if praestolatus came to be univerbated, it must have still been construed as a past participle. Therefore, the present of a verb back-formed from this would have to mean ‘bring oneself [so as to be] praesto ’, while praestolo(r) seems to mean simply ‘be praesto ’.6 In short, this theory encounters obstacles at every stage. Although fero is extremely common, it is apparently never found with praesto. The participial forms from which praestolor is said to have been formed, are virtually non-existent in older Latin. Thus there does not seem to have been sufficient time to completely lose these traces of the history of a univerbated praestolatus that was formed (or analogically re-formed) recently enough to show simple latus , rather than *tlatus or some development thereof. If praestolo(r) originally meant ‘be praesto ’, Stowasser’s idea of a derivation from the adverb is attractive in this respect at least. The difficulty, of course, is to account for the -l- of the stem. Some scholars have tried to solve the problem by denying the connection with praesto (adv.) entirely and proposing a root with a final -l- . Bre´al,7 following a suggestion of Meillet,8 argues that praestolo(r) is from the root *stel- ‘to place’, although, unlike Meillet, he may have believed that the medial -o- of praestolo(r) is short.9 Bre´al tries to support this etymology on semantic ' low ‘(sea-)voyage; expedition; fleet’ and grounds by comparing the use of sto ' stellv ‘rig out (a ship); (med-pass.) set out’ in nautical contexts, and therefore proposes that praestolo(r) properly means ‘wait on dry land (for a ship)’.10 There are two problems with this theory. Firstly, the long -o- of praestolo(r) is guaranteed at Plautus Epid. 221 at least. Secondly, the passages said to show that the verb could be used of people awaiting ships are neither so numerous nor so early as to be convincing examples of the original meaning of the word. The instance from Terence is especially puzzling since Parmeno has just encountered his master returning from the country to town; it is not clear that any boat is involved. Since all attempts so far to find a root in final -l- for praestolo(r) have been unsatisfactory, and since no objection can be made to the semantic appropriateness ¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Tragedies 29

"

praestolaras ¯ ¯

63

¯

of a derivation from praesto, it would seem best to begin from the adverb. First, however, it will be useful to consider the morphology of praesto (adv.) itself. Solmsen (Glotta 3, p. 245 ff.) suggests that praesto is the nominative singular of an on-stem, meaning ‘one who stands before’, that has become isolated.11 From this a denominative verb *praestona- ‘to be one who stands before’ ( 0/ ‘to await’) was formed, cf. caupo ‘tradesman’ : caupona- ‘to engage in trade’. This *praestona- then underwent a dissimilation of -t- . . . -n- to -t- . . . -l- , giving praestolo(r) . To grant the first premise of this theory, that praesto (adv.) is a fossilized substantive, one would have to accept that the rest of the paradigm was lost, but this is not a major obstacle; the same sort of thing must have happened according to almost any explanation of the adverb. The real difficulty emerges in the final stage, the dissimilation of -n- to -l- . No reasonably plausible parallels for such a development have been identified in Latin. A better explanation of praesto (adv. /It. presto ‘soon, quickly’) is suggested by the discussions of Prellwitz12 and Persson.13 The adverb would appear to be either the instrumental or ablative singular of a compound going back to *prai' -sth2 -o- , the second member being the zero-grade of the root of sto, and therefore meaning, in the first instance, ‘standing before’ or the like. Such a compound would give a Latin adjective praestus (/Fr. preˆt ‘ready’), and this is attested (CIL 6.12013.11), but may well be a back-formation from praesto (adv.) itself.14 Even if the attested adjective is not directly inherited, it must have recreated what had already existed, since the adjective is required to explain the adverb praesto.15 This analysis of praesto (adv.) is not entirely without problems either. If it is the instrumental or ablative of an o -stem, the -o should be old. But Cassiodorus the grammarian remarks: praesto nos per o scribimus, veteres per u scripserunt , sed sic praesto dicendum est ut sedulo (VII.157.22). The form praestu is even attested in inscriptions (CIL 6.4416; 6.37763 a & b), although apparently not from an early date, the first instance being dated probably to the Augustan period. If praestu¯, presumably with a long -u¯, is actually the original form of the adverb, it is probably from a u-stem substantive derived from the o -stem adjective, standing to praestus in the same relationship as promptu¯ to promptus . But this suggests another explanation for praestu¯, namely that, since praesto and promptu¯ are so similar in meaning, it is conceivable that praesto was remade to praestu¯ under the influence of promptu¯.16 In any case, there is no reason to believe that praestu¯ is the earlier form of the adverb. H. Ehrlich, however, dismisses Persson’s theory of praesto (adv.) as unsatisfactory, because it fails to take into account the formation of praestolo(r) .17 Ehrlich begins his own explanation of the verb by declaring his belief, on the basis of the Cassiodorus passage, that praestu¯ is the older form of the adverb18 and goes on to deny that the adverb praesto is in any way connected with the verb(s) praesto. From the fact that the adverb means virtually the same as praesens , Erhlich argues that prae-s-tu¯, ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

64

A Linguistic Commentary on Livius Andronicus

as he divides it, is the ablative of a verbal noun, of the type of portus and transitus , made from the weak stem of the root *h1 es- ‘to be’. To this he compares the ' w ‘absence’. Praestolo(r) , Ehrlich suggests, comes from a Hesychius gloss a'p-estu *prae-sto[u' ]-lo- , which he proposes is a proto-Italic derivative of the stem of prae-s-tu . Ehrlich’s pre-form *prae-sto[u' ]-lo- is both phonologically and morphologically difficult. The brackets around the *-u' - seem to mean that the glide disappeared when the *-lo- suffix was added. While this treatment of a *-u' l- cluster is not inconceivable, there seems to be no source for the preceding -o- in a derivative of a tu -stem. Also, one might rather expect that the *-ou' - would have been treated as a diphthong and come out eventually as -u¯-. Also, the morphological analysis of the -lo- suffix is unclear. It would appear that *prae-sto[u' ]-lo- is meant to be a diminutive in *-lo- . But Latin seems not to have any derivatives of this type from tu -stems. In short, the idea of a tu -stem seems altogether unlikely here and Ehrlich’s arguments do nothing to confirm that praestu¯ is older than praesto. Therefore, there is no reason to dismiss the idea that praesto is from an o -stem paradigm. Although Persson does not attempt to explain the morphology of praestolo(r) , there are at least two ways in which the verb could have been formed from an original praesto with the sort of history he proposes. First, if the adverb is an ablative, the inherited ending would have been *-od . From this *prai' stod, a first conjugation verb could be derived in the same way as autumare and negare are derived from the adverbs autem and nec respectively. The -d- in this *prai' stod-acould then have undergone a change to -l- .19 This would be the so-called ‘‘Sabine -l- ’’ also used to explain lingua (vs. Old Latin dingua and Gmc. *tungon- ), levir (vs. Gk. daÆr and Skt. devar- ), oleo (vs. odor ), etc. It is dangerous, however, to start from a *prai' stod , since the adverb seems never to be attested with a final -d , nor does it occur in hiatus with any noticeable frequency, if ever, in early Latin. But another explanation is possible that does not require praesto to be an old ablative. If a morphological boundary cannot be placed just after the -l- , then perhaps one can be placed before it. Ernout-Meillet propose, with a query, that praestolo(r) may be a denominative verb from a *praesto-lo- . Although they do not make it explicit, this pre-form is presumably meant to be a diminutive in -lo- from praesto (adv.). While diminutive forms of adverbs are known, e.g., clanculum (clam ), no such form from praesto is actually attested. We have already observed that Latin is capable of deriving verbs from adverbs (autumare, negare ) by simply adding -a- as the stem vowel. If one wanted to create such a verb from praesto, there would be a problem with the -a- contracting with the final vowel of the adverb20 and the result would probably be another verb praesto to add to the confusion of the other two. Latin can, however, form denominative verbs with the suffix -(e)la-, as well as with simple -a- , e.g., violare beside v is , for which there is no diminutive substantive attested.21 And although the -(e)l- of this ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Tragedies 29

"

praestolaras ¯ ¯

65

¯

formant is very probably to be traced to the inherited suffix *-e-lo- , which derives endocentric substantives from substantives, such a diminutive substantive is by no means a prerequisite for the creation of a denominative verb in -la-. Therefore the verbal suffix -la- would be a convenient way of avoiding this problem and keeping the adverbial basis separate from the verbal morphology. Another possible explanation of praestolo(r) is suggested by the fact that although in most cases praestolo(r) does seem simply to mean ‘wait for’ with no movement on the part of the subject indicated, there is at least one example in early Latin in which the verb would better be translated ‘go to meet’. In Plautus’ Epidicus , the title character comments on the great number of meretices in the city who are running to meet (obviam occurebant ) their lovers and describes how he has seen one particular woman: atque ego illam illi video praestolarier j et cum ea tibicinae ibant quattuor (217 f.). If four flute-players were going (ibant ) along with this woman, she is unlikely to have been standing still waiting. This raises the possiblity that praestolo(r) contains a verb of motion. Latin preserves a likely candidate in palor ‘wander’ ( B/*pe-al- ) and ambulo ' omai ‘walk’ ( B/*amb(i)-al- ), both compounds of a root *h2 el- also found in Gk. a'la 22 ‘wander’. Praestolo(r) would then have to be explained as a univerbation of this verb *alo and a form praesto meaning not ‘at hand, at the ready’, but rather ‘to hand, to the ready’. The semantics in that case would seem to require a directive case form. Remains of the inherited directive are few in Latin and mostly from pronominal stems, e.g., adeo ‘to this point’ and quoad ‘to what point’, whose function is clear from the univerbation with ad . But it is difficult to know whether the original directive of the o -stem adjective praestus would have been praesto. There is comparative evidence, however, to suggest that the inherited thematic directive in *-o-h2 e gave *-oh2 then *-o (/Hitt. -a), which would give -o in Latin.23 Nevertheless, it is not necessary to claim that praesto really preserves the original directive morphology. There was apparently a synchronic process in Latin by which functional directives were created by simply dropping the final -d from the corresponding ablative singular form; this is needed to account for expressions like mea re fert ‘it concerns me’. This process could easily have been suggested by pattern of the original directive quo ‘whither’ beside the ablative quod . Thus, from the ablative of a substantivized neuter of praesto- , meaning ‘readiness’ or the like, a new directive praesto ‘to the ready’ could be formed. If praestolo(r) does contain a directive and originally meant something like ‘come, go to the ready’, one would have to assume a slight semantic shift to ‘be ready’, but this would not be unnatural once the simplex *alo was lost and praestolo(r) was synchronically associated with praesto ‘at hand’. It seems most likely, therefore, that praestolo(r) is composed of a directive praesto, whether this was formed with the inherited directive morphology or created ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

66

A Linguistic Commentary on Livius Andronicus

later from the ablative of *prai' -st h2 -o- , and a verb *alo, probably also preserved in palor and ambulo. ¯

¯

¯

NOTES 1. Das Verbum lare , p. 14. 2. For ferr i¯ meaning ‘to proceed, go’, see meaning 4 under fero¯ in the OLD . The usage is attested as early as Pacuvius (Inc. 31 W). 3. There is but one instance of a perfect participial form in Plautus, at Truc. 336 (nescio quem praestolata est; credo militem ), although even here the text is not absolutely secure. There are no such participial forms in Livius, Naevius, Ennius, Accius, Pacuvius, Caecilius, Lucilius, Cato, or Terence. 4. On the possibility that praesto¯ (adv.) is an old abl., see below. 5. Note that this is a form of the verb ‘to be’, with which the adverb, appearing most commonly in the expression praesto¯ est (often written praestost ), is typically used. 6. This is essentially the definition of Festus (Praestolari is dicitur qui ante stando ibi , p. 223 M) and of Donatus (praestolari est praesto esse , on Eun. 975). 7. MSL 15, p. 141 f. 8. MSL 9, p. 15. 9. He comments ‘‘Le seul dictionnaire qui, a` ma connaissance, ait bien de´compose´ ce mot, celui de Vanicˇek, marque cependant l’o du signe de la longue. ’’ (p. 142, n. 1). 10. Bre´al cites Caes. B.C. II.23; Ter. Eun. 975; Cic. Att. II.15 as examples of such a meaning in Latin. 11. See also F. Skutsch, Glotta 2, p. 389 ff. 12. BB. 19, p. 318. 13. Beitr. 240 f. 14. cf. se¯dulus from se¯dulo¯ ( B/se¯ dolo¯). 15. Either praestus or praesto¯ is the basis of the verb praesto¯, -a¯re , -a¯v i¯ ‘make available, provide, offer’, a factitive of the (re -)nova¯re- type that has become confused with praesto¯, -a¯re , -it i¯ ‘be superior’, which is simply a compound of prae and sto¯. 16. F. Skutsch, op. cit. , p. 389, n. 1. 17. Berliner Philologische Wochenschrift (1913), p. 1201 f. 18. It is worth bearing in mind that even if we believe the ‘veteres ’ used a certain form, this is no guarantee that it is the older form. After all, Priscian (II.208.18) tells us that the ‘vetustissimi ’ used the form carnis as nominative singular, as in Livius Od. 45 W, instead of caro¯, which is clearly the inherited form. Not all innovations, however early the date, catch on. 19. See W-H under praesto¯lo (r ). 20. If the adverb is an old ablative, one would have to assume that the verb was created only after the final -d was lost. 21. For a more detailed discussion, see the appendix. It is also possible that viola¯re was formed from the ablative v i¯, used as an adverb. 22. See M. Weiss, Studies in Italic Nominal Morphology , Cornell University dissertation (1993), p. 53. 23. Melchert, p. 51, with reference to Jasanoff.

38 nefrendem, lacteam inmulgens opem tr. 38 W: quem ego nefrendem alui lacteam inmulgens opem Nefrendem is the accusative singular of an adjective whose nominative is unattested, but generally agreed to be nefrens B/*ne-frend-s . The word otherwise only occurs in contexts where it is being defined, but in this fragment, it is used of an infant nursing and therefore the meaning ‘not biting’, i.e., ‘toothless’, reported by Festus seems appropriate (ait Q . Mucius Scaevola esse arietes, quod dentibus frendere non possint; Ateius Capito infantes esse nondum frendentes , id est frangentes ).1 Varro, although referring to pigs, gives a very similar definition: amisso nomine lactantes dicuntur nefrendes ab eo, quod nondum fabam frendere possunt , id est frangere.2 The infrequency of the word led to some confusion already among the ancient grammarians. Festus adds that: sunt qui nefrendes testiculos dici putent and pro nefrendibus alii nefrundines intellegunt, quos usus recens dicit vel renes vel testiculos, ' w, Praenestini nefrones .3 Both of quos Lanuvini appellant nebrundines, Graeci nefrou these passages point to the error made explicit in the second, namely that nefrendhas become confounded with nefrundin-.4 It is nefrundines (B/*neg u' hro-n/d- ), ' w ‘kidney’ and which may be a dialect form, that is actually cognate with Gk. nefro Praen. nefrones ‘kidneys’. Nefrens would appear to be a compound, with the semantics of an agent, whose first member is the privative ne%- and whose second member is a root noun from the same root as frendere/frendere ‘to bite, gnash the teeth’ (*g u' hre(n)d- /Gmc. *grind‘grind’). The difficulty, as Bader points out,5 is that the full-grade version ne%- should not appear in a true nominal compound. Apparent cases, such as nefas ‘not right, wrong’, nescius ‘not knowing’, and nemo ‘no one’, are in reality univerbations, extracted from sentences like ne fas est . . . ‘it is not right . . .’, which have been reinterpreted as negative determinative compounds. Genuine compounds, on the other hand, regularly have the zero-grade form of the privative *n - (/Lat. in- , Gk. ˚ a ' -, Gmc. un- , etc.). In this line of Livius, *infrens would scan just as well and this raises the question, why would nefrens have been preferable? One can only speculate, but it seems possible that *infrens might have been misconstrued as containing the preposition/preverb in- , and hence thought to mean ‘biting (on)’, cf. infrendere . ¯

¯

¯

¯

¯

¯

67

68

A Linguistic Commentary on Livius Andronicus

A comparable alternation may be observed in the pair infans and nefans. Infans ordinarily means ‘not speaking’, but is attested once in Accius with the apparent meaning ‘not to be spoken (of )’.6 Nefans , on the other hand, is used (twice by Lucilius7) only in the latter sense, synonymous with nefandus. In any case, nefrens should not be a very old form and would appear to be analogical. Latin has very few words like nefrens with a stem of the shape -Vnd-, but one is deprans (Naev.) ‘not eating ( /not having eaten) lunch’. Furthermore, in deprans , the preposition/preverb de- has essentially the same privative sense as the ne- of nefrens, cf. deplumis ‘having no feathers’, etc. Therefore deprans , beside the verb prandere ‘to eat lunch’, would appear to be a likely model for the creation of a compound like nefrens from frendere . The basis of deprans itself may well have been an agent noun *prans (B/*pram-ed-s ) of the same structure as, for example, praeses (B/*prai' -sed-s ). If a reasonably old form *prans actually existed, it could provide an explanation for the verb prandere , whose -e- is otherwise difficult to account for, in the face of the simplex ede%re ‘to eat’.8 The pattern of praeses : praesidium : praesidere could have prompted the creation of prandere beside *prans and prandium . The formation of deprans is probably modeled, in turn, on compounds of the type l ibripens , lit. ‘balance-hanger’, (XII Tables , 8.22 W), whose second member is a root noun *pens B/*p(e)nd-s (cf. pende%re , pendere ) that also comes to have a stem ending in -Vnd-. *** The syntax of the phrase lacteam inmulgens opem is somewhat surprising at first glance. The verb means to ‘to milk into, to express milk into’. The thing into which the milk is expressed appears in the dative case, as labr is does in the following passages: ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

hic natam in dumis interque horrentia lustra armentalis equae mammis et lacte ferino nutribat teneris inmulgens ubera labris . (Verg., Aen . 11.570
/2), fabulosum enim arbitror de strigibus ubera eas infantium labris immulgere . (Plin., Nat. 11.232). But in both of these examples ubera is in the accusative, whereas in the Livian fragment the milk itself, lacteam opem , is in that case. There seems to be no other instance in classical Latin of ‘milk’ as the direct object of inmulgeo. Only very late, starting in the fourth century AD, do we find parallels to the Livian syntax, first in Firmicus Math. (nato homini . . . inmulgentur alimenta nutricia, 3.14.10) and later yet in Symmachus (praecepta rhetoricae pectori meo senex . . . inmulsit, epist. 9.88.3). ¯

¯

Tragedies 38

"

nefrendem, lacteam inmulgens opem

69

In the one other place where Pliny uses inmulgeo the verb is passive but its subject cannot be the ubera one might expect from the active usage: ¯

¯

. . . oculo ictu cruore suffuso et in dolore aut epiphora , si inmulgeatur , plurimum prodest , (Nat. 28.72). The passage can be translated: ‘‘[milk] is most beneficial for an eye bloodshot and painful from a blow or an inflammation, if milked into [the eye].’’ If ‘milk’ can be the subject of a passive form of inmulgeo, it follows that an active form of the verb could take ‘milk’ as an accusative object. This passage could be indirect evidence that Livius’ construction lacteam inmulgens opem is not so anomalous. The phrase si inmulgeatur , plurimum prodest is open to another interpretation, however. If inmulgeatur and prodest are being used impersonally here, rather than with lac as the subject, the sentence could then be translated: ‘‘it is most beneficial . . . if there is milk put in.’’ Therefore, this passage of Pliny is probably too ambiguous to be useful for these purposes. Thus we are left with inmulgeo indicating an action properly performed only on ubera . But there are expressions in Latin in which the word for an item that is pressed is replaced by the word for the product obtained by that pressing. For example, we find phrases like ‘‘oleum premi oportebit ’’ (Col. RR 12.52.22), literally ‘the oil ought to be pressed’, although of course what is actually to be pressed are the olives (oleae or olivae ) from which the oil is extracted. In just such a way, Livius may be substituting lacteam opem ( /lac ), the result of pressing ubera , for the ubera themselves. ¯

¯

¯

¯

¯

NOTES 1. 2. 3. 4. 5. 6. 7.

P. 163 M. RR 2.4.17. P. 163 M; cf. p. 277 M.: rienes quos nunc vocamus , antiqui nefrundines appellabant . See also Bader, p. 61. P. 37. tr. 156 W: prius quam infans facinus oculi vescuntur tui . 3.136
/7 W: Tantalus qui poenas , ob facta nefantia , poenas pendit . 21.886: . . . dissociata aeque omnia ac nefantia . 8. The old pres. inf. act. is, of course, esse .

Conclusion

The fragments of Livius Andronicus are notable for preserving inherited archaisms, such as the a-stem genitive singular in -as , but this study has shown that there is a surprisingly high degree of innovation within Old Latin. Several features that have been taken as archaisms are demonstrably innovated. The i' o -stem vocative singulars like f ilie do not directly continue an inherited *-ii' e , since this had become - i already in Italic. Rather, -ii' e has been analogically ‘‘restored’’ in Latin (and in Umbrian) by analogy to the o -stems (nom. sg. -o-s : voc. sg. -e : : nom. sg. -ii'o-s : voc. sg. X /-ii' e ). It has been argued that the third plural present endings -nunt (ur ) began from an archaic paradigm of ‘to go’ (preserved in ob inunt , prod inunt , and red inunt ) connected with Lith. einu` ‘go’ and/or Hitt. iyannai- ‘start moving’. But this is unlikely given the restriction of the -n- to the third plural in Latin and the limitation of the forms almost exclusively to archaic poetry. The -nunt (ur ) forms are better explained as an inner-Latin development starting, not from - inunt , but from danunt , which was created as a by-form of dant by the analogy, situs : sinunt :: datus : X /danunt . Homines has been emended in Od. 33 W to homones, the latter often thought to be the older form of the stem (cf. Enn. homonem ). But the forms cited as evidence of an Italic on -stem (Osc. humuns and Umbr. homonus ) could reflect -o%n- as well as -on-. Homon- seems rather to have been analogically created within Latin beside older *homo%n- on the model of a growing number of stems in -on- . Therefore it is unnecessary, and probably undesirable, to read homones in Livius. Nevertheless, the on -stem by-form was an innovation of early Latin. Finally, it is striking that none of these particular innovations survived; in each case the Classical form is more archaic than what is found in Livius Andronicus. ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

71

Appendix: The origin of the Latin adjectives in -ulentus

The origin of the adjectives in -ulentus is a long-standing problem of Latin morphology. No inherited suffix could have given Latin -ulentus directly and therefore most scholars have sought an origin for the formant within Latin itself. Perhaps the best discussion of the type is that of Szemere´nyi,1 whose investigation focuses on those of the -ulentus adjectives that are attested relatively earlier, before c. 50 AD. This cut-off date is determined by the pattern of the attestations themselves; roughly half are first found before c. 50 AD, after which there seems to have been a period of non-productivity until about the middle of the second century when the other half begin to appear.2 The earlier attested forms3 are: aquilentus ‘watery’ (?) (Var.) ancunulentus ‘‘-ae feminae menstruo tempore appellantur’’ (Paul. Fest .) bucculentus ‘having fat cheeks’ (Pl.) corpulentus ‘of a heavy build of body, corpulent, large’ (Pl.
/) esculentus ‘suitable for food, eatable; (?) eating-vessels’ (Scaev.
/) faeculentus ‘full of sediment, thick; cloudy’ (Col.
/) fraudulentus ‘dishonest, deceitful, fraudulent’ (Pl.
/) frustulentus ‘full of crumbs or morsels’ (Pl.
/) gracilentus ‘slender, thin’ (Enn.
/) iurulentus ‘containing juice or gravy, stewed’ (Var.
/) lotiolentus ‘defiled with urine’ (Titin.) ¯

¯

73

74

A Linguistic Commentary on Livius Andronicus

luculentus ‘excellent, splendid, fine’ (Pl.
/) lutulentus ‘full of mud, muddy, miry’ (Pl.
/) macilentus ‘thin, lean’ (Pl.
/) mustulentus ‘full of unfermented wine’ (Pl.
/) obstrudulentus ‘that can be swallowed down’4 (Titin.) opulentus ‘having many possessions, wealthy, opulent’ (Pl.
/) pestilentus ‘insalubrious, unhealthy’5 pisculentus ‘teeming with fish; made of fish, fishy’ (Pl.
/) posculentus ‘drinkable’ (Scaev.
/) pulverulentus ‘covered with or full of dust, dusty’ (Var.
/) purulentus ‘containing or consisting of much pus, purulent’ (Cato
/) rorulentus ‘wet with dew’ (Cato
/) sanguinolentus (-ul- ) ‘covered with blood, blood-stained, bloody’ (Quad.
/) temulentus ‘drunken’ (Ter.
/) truculentus ‘ferocious, aggressive’ (Pl.
/) turbulentus ‘violently disturbed, stormy, turbulent’ (Pl.
/) v inolentus ‘immoderate in one’s consumption of wine, intoxicated’ (Pl.
/) violentus ‘acting with force toward others, violent’ (Pl.
/) ¯

¯

¯

¯

¯

¯

Szemere´nyi offers what seems to be now the most generally accepted explanation, namely that -ulent- is dissimilated from *-en-ont- . Before turning to this, however, it may be useful first to discuss briefly the inadequacy of some of the previous attempts to explain -ulentus . Ernout6 argues that -ulentus was an entirely Latin creation made by combining the -ul- of the type of adjective exemplified by credulus ‘trusting’ and bibulus ‘drinking, thirsty’ with an element -ento- . This -ento- he proposes to have been extracted from cruentus ‘bloody’. Leumann7 had earlier proposed that cruentus was responsible for part of -ulentus , but it is not clear exactly how. As others have noticed, his proportion, cruentare : violare : : cruentus : X, does not yield violentus . But Ernout, likewise taking violentus as the point of departure for the whole class, hypothesizes an adjective *vi-ulus of the credulus-type as the basis of the verb violare . He suggests that the semantic similarity between this *vi-ulus and cruentus could have encouraged the blending of their suffixes to give violentus . How likely is it, though, that a single, sychronically unanalyzable word like cruentus should be fixed upon as a source of a new suffix for a whole class of derived adjectives? Also, it is not clear what a combination of this suffix -ento- and an adjective with the semantics of an agent, like credulus , would mean. In fact, few of the -ulentus adjectives are analyzable as expansions of any sort of adjective; there are only gracilentus beside gracilis (and gracilus ) and truculentus beside trux . ¯

¯

¯

¯

¯

¯

Appendix

75

Another objection to this scenario is that credulus and bibulus are deverbatives made from the present stems of credo and bibo respectively, whereas Latin has no stem from which a *vi-ulus could have been derived. Also, the credulus-type of adjective is not very common in Latin and not a single verbal stem provides both an adjective of this type in -ulus and one in -ulentus . In fact, virtually all of the -ulentus adjectives, with the exception of obstrudulentus and perhaps ancunulentus , would appear to be made to nominal stems in the first place. The theory of A. Zimmermann8 takes advantage of the other, much larger, group of Latin words in -ulo- B/*-e-lo- , namely the diminutives of the type of porculus ‘small pig’ from porcus. He proposes the analogy gracus (Lucil., Ter.) : gracilus : : gracentes (Enn.) : X /gracilens . Notice that this explanation requires that the athematic by-forms in -lens be the older, which is far from certain, and may even qualify as unlikely. It may be useful at this point to discuss briefly the evidence in this question of whether -lens or -lentus represents the earlier inflection.9 If we consider from what dates the -lens forms are actually attested, they certainly seem to be a later development; violens first appears in Horace while violentus is known from Plautus. Similarly, Plautus uses opulentus , but opulens is first attested in Sallust. On the other hand, gracilens (Laev.) and gracilentus (Enn.
/) are both found in early authors. And pestilens is found as early as Cato, whereas pestilentus may not really exist at all, since the only potential instance of it is in the phrase pestilenta loca , which is completely ambiguous. The appearance already in Plautus and Terence of the derived adverbs of the type opulenter 10 could be taken as further evidence that the -lens forms are original. But these adverbs in -ter could easily have been analogically created at any time without the prior existence of the athematic adjective. In short, it is not probable that the -lens forms are the older. But Zimmermann’s analogy is open to a more decisive objection. If we consider all the -ulentus adjectives attested from about 50 AD or earlier, it is striking how few of them actually occur beside diminutives. Beside faeculentus (Col.
/), there is faecula (Lucr.
/) and beside bucculentus (Pl.), buccula is given in one manuscript at Plautus Truc. 290.11 In other cases, however, even when the diminutive exists, it is first attested at a significantly later date than the -ulentus adjective which is supposedly derived from it; for example, whereas frustulentus and turbulentus are known from as early as Plautus, frustulum and turbula first appear in Apuleius.12 In favor of a derivation from an earlier diminutive, one could add that turbella (Pl.
/) is at least as old as turbulentus , and likewise macellus (Lucil.
/) beside macilentus . But although, from a historical point of view, diminutives in -ello- are the result of essentially a doubling (*-el-el-o ) of the same suffix that gives -ulo- , these particular diminutives in -ello- do not necessarily presuppose earlier diminutives in -ulo- from the same stems. Turbella , for example, could have been ¯

¯

¯

¯

¯

¯

76

A Linguistic Commentary on Livius Andronicus

created analogically from turba at any time once the pattern porc- , porcul-, porcellhad become established. There would have been no need for an intermediate turbula stage.13 In still other cases, the diminutive is made in a different way altogether, so beside corpulentus we have corpusculum , not *corpulum , and beside pisculentus, pisciculus , not *pisculus . A special case is aquilentus. While it is true that aquola or acula (Pl.
/) is older than aquilentus , if this diminutive were the basis for the creation of the adjective, one would have expected *aquolentus or *aculentus . If aquilentus is derived from the diminutive, it must have undergone some further modification. But it is not even clear if aquilentus has anything to do with aqua. The OLD gives the definition ‘watery’, but the only instance of the adjective is in Varro, who uses it to describe the moon.14 Aquilentus , and the other words with the variant -ilentus, will be discussed further soon. In short, -ulentus resists all attempts to break it down into a concatenation of one or another of the suffixes in eventual -ulo- and an -ento- of some origin. Therefore, taking a different approach, Szemere´nyi came to the conclusion that -ulent- is the result of a dissimilation from an inherited *-en-ont- .15 His prime example is opulentus itself, which he analyzes as a thematized version of an IE *op-en-ont- , a denominative adjective which could also be the pre-form of Hitt. h% appinant- ‘rich’. Szemere´nyi’s analysis may be questioned at several points. Firstly, Hitt. h% appinant- may in fact be a derivative in -ont- from the r /n-stem preserved in Hittite,16 but it could also be the adjective h% appina- ‘rich’ with an added *-nt- , a type of expansion process that is well-known in Hittite, as Szemere´nyi himself mentions.17 Although this sort of addition to an adjective is common in Hittite, it is */to say the least */not at all a wide-spread phenomenon in Latin. Secondly, most of the examples given for the dissimilation of sequences of -n- . . . -n- to -l- . . . -n- date from the time of Romance (Bononia 0/ Bologna , Barcinona 0/ Barcelona , venenum 0/ It. veleno , Saturn inus 0/ Fr. Saint-Sorlin ), rather than from classical Latin or earlier. Szemere´nyi also cites as a parallel the name Capitol inus , which Niedermann18 proposed to be dissimilated from *Capiton inus . At best, this example is ambiguous; it is difficult to see why *Capitol inus must be a dissimilation, rather than simply a derivative of Capitolium (which cannot have its -l- from such a dissimilation), like Lat inus from Latium . On the other hand, lympha , in one of its meanings at least, is probably from Gk. ' mfh, although it may not have been borrowed directly from Greek into Latin. nu Even if acceptable, however, as an explanation of opulentus , this dissimilation is not convincing as the origin of the entire class, since opulentus is the only one of the type for which a reconstruction *-en-ont- is at all likely. Szemere´nyi suggests that ¯

¯

¯

¯

¯

¯ ¯

¯

¯

¯ ¯

¯

¯ ¯

¯

Appendix

77

turbulentus may be derived from turbo, turbinis , rather than from turba , but this example is, of course, completely ambiguous. Szemere´nyi did recognize that the main obstacle to any explanation of -ulentus is the source of the -l- . If it is not from the credulus -type of adjective, on the one hand, nor from the porculus-type of diminutive, on the other, then where did it come from? Let us approach the problem now from a slightly different angle. Leumann and Ernout had tried to exploit the fact that one -ulentus adjective, namely violentus , is unique in being connected with a similarly problematic verb, violo. But the attempts to derive this verb from a *violus or *viola are no more convincing than the proposals to get violentus from the same sorts of pre-forms, and for the same reasons. A nominal form in *-lo- , however, is not a prerequisite for a verb in -lare . Although the ‘‘diminutive’’ suffix *-e-lo- originally derived endocentric substantives from substantives, as described above, the resulting -ulo- in Latin was extended to adjectival bases as well, including participles, e.g., edentulus , valentulus , and blandiloquentulus . Eventually, the -l- of this suffix seems to have become synchronically perceived as a mark of diminution in itself. This led to its use in the formation of what are sometimes called ‘‘diminutive verbs’’, although in many cases the diminutive force is no longer perceptible. Historically speaking, these seem ultimately to have the -l- of the diminutive nouns, but can be either denominative, like ustulo ‘singe’ (ambustulatus , Pl.) from ustus ‘burnt’ and testilor (a variant of testiculor according to Festus) from testis , or deverbative, like obvagulo ‘wail at’ (obvagulatum , XII Tables ) from vagio ‘wail’ and petulans from peto.19 Therefore, violo may be described as a present with the ‘‘diminutive’’ denominative formant *(-e)-la- in place of simple *-a-. It is unnecessary to presuppose a diminutive noun, which does not survive, as the basis of the verb. Latin also has another set of -l- forms that are of neither the credulus - nor the porculus-type. These are adjectives in which, descriptively speaking, the suffix -ulois a (presumably diminutivizing) substitute for -uu' o- . In Latin, the suffix -uu' omakes deverbative adjectives, with active force when the verb is intransitive, as in occiduus ‘sinking, setting’ (lit. ‘falling down, cf. cado) and innocuus ‘harmless, innocent’ (lit. ‘not harming’, cf. noceo), but with passive force when the verb is transitive, as in perspicuus ‘clear, transparent’ (lit. ‘seen through’, cf. specio) and exiguus ‘meager’ (lit. ‘forced out’, cf. ago). Whatever their remoter history, delicuus ‘falling short, failing’ (Pl.
/) is an adjective of this type and we find beside it deliculus ‘imperfect, blemished’ (Cato), both clearly derivatives of the intransitive verb (de- )linquo ‘to fail, be wanting’. But deliculus is at least assumed to have a slightly hypocoristic meaning*/the OLD defines it as ‘having a (small) blemish’*/ and certainly looks as if it has been formed by simply replacing -uu' o- with -ulo-. ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

78

A Linguistic Commentary on Livius Andronicus

The same process would account for iaculus ‘used for throwing’ (Pl.).20 One would expect such a deverbative to be made with the same -uu' o- as the examples above. But there is no *iacuus attested, so it would seem to have been replaced completely by the ‘‘diminutive’’ version parallel to deliculus . Likewise, reiculus ‘discarded’ (Var. ap. Non.
/) has taken the place of *reicuus .21 Thus it seems that the type represented by deliculus , iaculus , and reiculus further exemplifies the spread of -l- in its capacity as a synchronic diminutive marker. Something slightly different, however, is going on with the pair cernuus (Lucil
/) and cernulus (Var.
/), both meaning essentially ‘with the head inclined’.22 In this case the -u' - apparently being replaced has another source ' entirely. Cernuus is best explained as a compound *ker h2 o-nou' o- , of the foedifragus type, with a first member exactly cognate with Av. sara- ‘head’ and a second member derived from the verb (ad-)nuo ‘to nod’.23 Notice that in cernuus the -u' belongs, not to a suffix, but to the verbal root itself. This case shows that the synchronic process was -uu' o- /-ulo- , whatever the source of the -u' - . From alternations like delicuus/deliculus and cernuus /cernulus, it is possible to get the impression that -l- could be substituted for *-u' - with little or no apparent difference in meaning. This raises another possibility for the source of the -l- of -ulentus : what if it is a replacement of a *-u' - ? On this assumption, the immediate pre-form of -ulent- would be *-uu' ent- . The next question, therefore, is how is this *-uu' ent- to be explained? Whereas -ulent- could not be equated with any IE suffix **-lent- , thereby causing scholars to try to divide it into an -ul- plus an -ent- , there is a perfectly good IE suffix *-u' ent- . If -ulentus goes back ultimately to an inherited *-u' ent- , then the thematization of the suffix must be secondary. Such a thematization of a stem in -nt- can, however, be paralled in Latin. Although the details of its reconstruction are quite complicated, cruentus ‘bloody’ is generally agreed to contain an inherited athematic suffix, either *-ont-/ *-ent- or perhaps even *-u' ent- , which has here been thematized.24 The only extra piece of morphology is the surface -u- or -o- that comes to precede the -lent- resulting from *-u' ent- . The few examples of an -i- in this position (e.g., gracilentus ) are demonstrably secondary and will be discussed below. The fact that -o- is found following an -i- , as in violentus , but -u- appears in most other environments, suggests that the -o- and -u- are the regular phonological developments of an original *-e- or *-o- in these positions. The only likely candidate is the thematic vowel, which frequently serves to link the two halves of compounds in Latin. Compare also the situation in Greek where an -o- has been introduced into precisely this formation during the historic period, giving, for ' eiw ‘bloody’ as early as Homer versus Myc. (fem.) pe-de-we-sa example, a•jimato 25 ‘having feet’. ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Appendix

79

Already in 1905, H. Ehrlich proposed that the -ulentus adjectives contained the suffix *-u' ent- .26 Again taking the by-forms in -lens to be the older, he set up preforms of the type *gracile-u' ent- and *viole-u' ent- which could contract to give gracilens and violens . But, as Szemere´nyi observed27, the long -e- preceding the suffix is morphologically inexplicable. That the -ulentus adjectives originated in a dissimilation of -u' ent- to -lenthas, in fact, already been suggested by Ribezzo in a review of Stolz-Schmaltz’s Lateinische Grammatik (5er Aufl. vo¨llig neubearbeitet von Manu Leumann u. J. B. Hofmann [Mu¨nchen, 1926]).28 He gives so few details, however, that Szemere´nyi dismisses the idea with the terse comment: ‘‘Nor can any reason be found why, or how, an original *w ino-went- should have been dissimilated to v inolent-’’.29 I hope I have shown how such a dissimilation could have come about; it only remains, therefore, to explain why. Even granted that it is possible for an inherited *-u' - to be replaced by -l- in Latin due to a morphologically conditioned analogy to the alternations described above, we must still ask what would have been the motivation, in this case, to change *-u' ent- to -lent- ? Looking again at the list of relatively early -ulentus adjectives, one can hardly help noticing that virtually all of them have roots containing a labial or labio-velar consonant, a u' -diphthong, the vowel -u- , or even some combination of these. Of the twenty-nine words in our list, twenty-one have a labial consonant, either visible on the surface or reconstructible for the immediate pre-form, and fourteen a u- vowel or u' -diphthong. Fully one-third of the list have two such labial elements. The following list shows the labial elements that could have triggered the dissimilation: ¯

¯

¯

¯

¯

-u' -: pulver- , v in-, vi-u% -: ancun- , bucc- , frust-, lut- , must- , pulver- , truc- , turb-u- : iur- , pur ' -u' -diphthongs (or developments thereof): fraud-, loti- (*lau' t- ), luc- (*leuk - ), obstrudlabio-velars (or reflexes thereof): aqu-, sanguin- , corp- (*k u'orp- B/*k u' r p- , ˚ cf. Gk. pre' pei), ancun- (if connected with inquino ; see E-M) -p- : corp- , op- , pest- , pisc- , posc- , pulver- , pur-b- : bucc- , obstrud- , turb-f- : faec- , fraud- , frust-m- : mac- , must- , tem- . ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

The exceptions are very few (esculentus , gracilentus , and rorulentus ) and mostly easily explicable, as we shall see momentarily. This is almost certainly not a ¯

80

A Linguistic Commentary on Livius Andronicus

coincidence. A suffix *-u' ent- , added to such a quantity of labial sounds, would surely have made these words likely candidates for a dissimilation. That a velar -l- should be an acceptable substitute for *-u' - is not too surprising, given that in Latin these two sounds pattern together in certain ways. Like -u' - , the l-pinguis causes what might be called rounding effects. It is a general rule of Latin that eu' /ou' and (even though it may be a later change) el /ol in all syllables. Also, in medial syllables, -u' - and l-pinguis are the only two consonants before which a short vowel is regularly reduced to a -u- , e.g., praesulis , etc. (B/*-sal- , cf. salio), depuvio ( B/*-p a% v-, cf. pavio). Typologically, this is as likely a dissimilation as -n- . . . -n- to -l- . . . -n- . The affinity between -u' - and -l- seems to be paralleled in Umbrian, which shows a change of *-l- /-u' - in, for example, Vuvcis ‘Lucius ’ and vutu ‘lavato’.30 In the Cretan dialect of Greek it is not uncommon to find -u- for -l-, as in adeupia¤ § for j ' w for kalxo ' w.31 We might also compare the development of adelfea¤ §j and kauxo -l- to -u- in, for example, Fr. autre from Lat. alter . Therefore, if a -u' - should undergo dissimilation at all, a velar -l- is a very likely outcome. The instances of -i- preceding the -l- would, therefore, have to have been secondarily altered. In fact, it can be shown that there is a fairly obvious source for the -i- in each of these words. Gracilentus could simply have adopted the -i- from its synonym gracilis , and macilentus , with its very similar meaning, could in turn be modeled on gracilentus , although macies may also have had some influence.32 Pestis and pestilitas may well have played a role in creating pestlilens. Aquilentus , however, is more problematic. It is used only by Varro to describe the moon and so it is difficult to tell what exactly it means. The usual gloss ‘watery’ is based on the assumption that it is derived from aqua . If this is the right meaning, and if aquilus ‘dark’ is somehow derived from aqua also, then the -i- may be from aquilus , although it would be irregular there too. The -i- may also be from aquilus , whatever its etymology, if we are willing to entertain the idea that aquilentus means something like ‘dark’ as well. But the relationship, if any, between these words has not yet been satisfactorily explained. To summarize, I would propose, as did Ribezzo, that in the first instance, perhaps in only a few words such as *u' i-o-u' ent-o- and *u' oi' no-u' ent-o- , the occurrence of the two -u' - sounds in rapid succession motivated a dissimilatory treatment. With a dissimilation thus motivated, the question was how exactly to differentiate one of the -u' - ’s. To accomplish this, advantage was taken of the fact that elsewhere -l- , ultimately traceable to the diminutive suffix, appeared to alternate synchronically with -u' - , without necessarily conveying a strong diminutive sense, as in cernuus/cernulus , etc. The ‘‘new’’ suffix *-olento- thus created then spread to other stems where the accumulation of labials may have been considered excessive. In fact, this suffix may ¯

¯

¯

¯

¯

¯

¯

¯

¯

Appendix

81

even be said to have favored stems characterized by labial elements in its initial period of productivity. I do not mean to deny that the suffix has spread to a considerable extent through the semantic similarities (such as in the group v inolentus, temulentus , mustulentus , posculentus) noticed by Leumann and others.33 In fact, this approach can explain the few instances where there was no problematic labial element; esculentus is only attested paired with posculentus and may easily be modeled on it, and gracilentus is synonymous with macilentus . The only real exception among the early forms is rorulentus. Of course, -ulentus eventually becomes quite productive and the spread to stems without labials had to begin somewhere. But it is surely significant that this is the only such stem we know to have made an -ulentus adjective until the end of the first century AD. Finally, if -lent- is really the treatment of *-u' ent- in these adjectives, we would have evidence that this familiar IE possessive suffix existed in Latin. Hence this theory has the additional advantage of answering the question of where the potential examples of *-u' ent- went. ¯

¯

¯

NOTES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

14. 15. 16. 17. 18.

‘‘The Latin Adjectives in -ulentus ’’ (Glotta 33 [1954] p. 266 ff.). Szemere´nyi, op. cit. , p. 266 /8. Definitions are from the OLD , unless otherwise stated. This is the definition of Lewis and Short; the OLD simply refers to the citation. This is only cited, however, in the phrase pestilenta loca , Laev. ap. Gell., so perhaps only pestilens (Cato
/) exists. Les adjectifs latins en -o¯sus et en -ulentus (Paris, 1949) p. 89 f. This theory seems to have disappeared from later editions of L-H; Szemere´nyi gives the reference as L-H 229. ‘‘Zu latein. Suffixen: 1. -u(i)lentus . 2. -os(s)us ’’ (KZ 44 [1911] p.13 ff.). The by-forms in -lens are: gracilens (Laev.), opulens (Sall.), pestilens (Cato
/), and violens (Hor.
/). Those that are attested reasonably early are: fraudulenter (Cato
/), te¯mulenter (Col.), lu¯culenter (Cic.), turbulenter (Cic.), opulenter (Pl.
/), violenter (Ter.
/). Bucculus also seems to be the title of a play by Novius (see Non. p. 507 M.). A similar case is that of lu¯culentus , which is Plautine, while the adverb antelu¯culo¯ appears first in Apuleius. As mentioned above, turbula is found, but only considerably later and may not have been created much before then. Beside macellus there seems to be no *maculus or *macilus at all. (Macula ‘stain, spot’ and macilis ‘dappled, piebald’ do not fit semantically with macilentus and are probably unrelated to it.) Men. 400. Lewis and Short interpret aquilentus as meaning ‘bringing rain’. Op. cit. , p. 275 ff. For a fuller discussion of the Hittite forms, see the comment on ops above. Op. cit. , p. 277. Cf. assu-/assuwant- ‘good’, irmala-/irmalant- ‘ill’. Me´l. Ernout (1940) p. 267 f.

82

A Linguistic Commentary on Livius Andronicus 19. Compare also the verbs in -culo(r) (gesticulor , etc.) and -illo(r) (conscribillo¯, etc.). See further L-H, § 414.4. 20. For this development of the passive meaning regular for transitive verbs, compare pascuus ‘used for feeding’ and caeduus ‘suitable, ready for felling’. 21. This adjective is also used in a substantivized feminine form to mean ‘a ewe or other animal culled out of a flock or herd on account of old age, etc.’ (OLD ). 22. Compare also cernula¯re ‘to throw headlong’, clearly a factitive of the (re)nova¯re- type from the latter adjective. 23. Nussbaum, p. 114. 24. See also Szemere´nyi, op. cit. , p. 279. 25. That is, pedwessa B/*-u' eti' a ; the consonant cluster -dw- is split between two syllabograms, the vowel of the first being a ‘‘dummy’’ determined by the vowel following the cluster. In Indo-Iranian and Hittite, the suffix is also added directly to the stem, e.g., Skt. padvant- ‘having feet’, Hitt. h% uisu' ant- ‘having life, alive’; this was no doubt the inherited situation. 26. ‘‘Die nomina auf -euw’’ (KZ 38 [1905] p. 53 ff.). 27. Op. cit. , p. 273. 28. RIGI 10 (1926), p. 293 ff. 29. Op. cit. , p. 274. 30. See Buck OU , § 104, and Untermann, p. 866 f. 31. See Buck GD , § 71. Thanks are due to Nina Loney for reminding me of the Cretan facts. 32. Watkins has pointed out that macilentus has an interesting correspondent in Hitt. maklant ‘thin’ (‘‘Hittite and Indo-European Studies: the denominative statives in -e¯- ’’ Transactions of the Philological Society [1971], p. 87). 33. See Szemere´nyi (op. cit. , p. 269) with references.

Bibliography

Bader, F. La formation des compose´s nominaux du Latin (Paris, 1962) Buck, C. G. Buck, C. D. Comparative Grammar of Greek and Latin (Chicago, 1933) Buck, O. U. Buck, C. D. A Grammar of Oscan and Umbrian 2 (Boston, 1928) Buck, G. D. Buck, C. D. The Greek Dialects 2 (Chicago, 1955) Buechner, C., ed. Fragmenta poetarum Latinorum epicorum et lyricorum praeter Ennium et Lucilium (Leipzig, 1982) Chantraine, P. Grammaire home´rique (Paris, 1948 ff.) De Vries, J. Altnordisches etymologisches Wo¨rterbuch (Leiden, 1961) E-M. Ernout, A. and A. Meillet. Dictionnaire e´tymologique de la langue latine. Histoire des mots (Paris, 1959) Evans, D. S. A Grammar of Middle Welsh (Dublin, 1964) Feist, S. Etymologisches Wo¨rterbuch der Gotischen Sprache (Leiden, 1909) Flobert, P. Recherches sur les verbes de´ponents latins (diss. Paris, 1973) Fraenkel, E. Litauisches etymologisches Wo¨rterbuch (Heidelberg
/Go¨ttingen, 1962) Frisk, H. Griechisches etymologisches Wo¨rterbuch (Heidelberg, 1960) Geldner, K. Der Rig-Veda (Cambridge, MA
/London
/Leipzig, 1951) Grassmann, H. Wo¨rterbuch zum Rig-Veda (Wiesbaden, 1955) Huld, M. Basic Albanian Etymologies (Columbus, OH, 1984) Jasanoff, J. Stative and Middle in Indo-European (Innsbruck, 1978) Jocelyn, H. D., ed. The Tragedies of Ennius (Cambridge, 1967) Keller, O. Lateinisches Volksetymologie (Leipzig, 1891) Klingenschmitt, G. Das Altarmenische Verbum (Wiesbaden, 1982)

83

84

A Linguistic Commentary on Livius Andronicus

Kronasser, H. Etymologie der Hethitischen Sprache (Wiesbaden, 1966) Lebek, W. D. Verba Prisca (Go¨ttingen, 1970) L-H. Leumann, M. and J. B. Hofmann. Lateinische Grammatik 5 (Mu¨nchen, 1977) Lewis, H. and H. Pedersen. A Concise Comparative Celtic Grammar (Go¨ttingen, 1937) LSJ. Liddell, H. and R. Scott, H. Jones, R. McKenzie. A Greek
/English Lexicon (Oxford, 1961) Lindsay, W. Early Latin Verse (Oxford, 1922) Lindsay, W., ed. Nonii Marcelli de compendiosa doctrina (Leipzig, 1903) Lindsay, W., ed. Sexti Pompeii Festi de verborum significatu quae supersunt cum Pauli epitome (Leipzig, 1913) Marinetti, A. Le iscrizioni sudpicene I: testi (Firenze, 1985) Mariotti, S. Livio Andronico e la traduzione artistica (Milan, 1952) Meiser, G. Lautgeschichte der Umbrischen Sprache (Innsbruck, 1986) Meister, R. Die griechischen Dialekte (Go¨ttingen, 1882
/1889) Melchert, H. C. Anatolian Historical Phonology (Amsterdam, 1994) Meyer-Lu¨bke, W. Romanisches etymologisches Wo¨rterbuch (Heidelberg, 1935) Monier-Williams, M. Sanskrit
/English Dictionary (Oxford, 1899) Morel, W., ed. Fragmenta poetarum Latinorum epicorum et lyricorum praeter Ennium et Lucilium (Leipzig, 1963) Nussbaum, A. Head and Horn in Indo-European (Berlin
/New York, 1986) OLD. Oxford Latin Dictionary (Oxford, 1982) Pauly-Wissowa. Real-Enzyclopa¨die der classischen Altertumswissenschaft (Stuttgart, 1894 ff.) Pedersen, H. Vergleichende Grammatik der Keltischen Sprachen (Go¨ttingen, 1909) Peter, H., ed. Historicorum Romanorum fragmenta (Leipzig, 1914) Pokorny, J. Indogermanisches etymologisches Wo¨rterbuch (Bern
/Mu¨nchen, 1959, 1969) Ribbeck, O., ed. Scaenicae Romanorum poesis fragmenta (Leipzig, 1897) Risch, E. Wortbildung der homerischen Sprache (Berlin
/Leipzig, 1937) Rix, H. Historische Grammatik des Griechischen (Darmstadt, 1976) Schmid, W. Studien zum Baltischen und Indogermanischen Verbum (Weisbaden, 1963) Schwyzer, E. Griechische Grammatik (Mu¨nchen, 1939 ff.) Sihler, A. L. New Comparative Grammar of Greek and Latin (Oxford, 1995) Skutch, O., ed. The Annals of Q . Ennius (Oxford, 1985) Sommer, F. Handbuch der Lateinischen Laut- und Formenlehre (Heidelberg, 1902) Stowasser, J. M. Das Verbum LARE (Vienna, 1892) TLL. Thesaurus Linguae Latinae (Leipzig, 1900 ff.) Thumb, A. and E. Kieckers. Handbuch der griechiscen Dialekte 2 (Heidelberg, 1932
/59) Thurneysen, R. A Grammar of Old Irish . tr. D. A. Binchy and O. Bergin (Dublin, 1946) Tischler, J. Hethitisches etymologisches Glossar (Innsbruck, 1977) Untermann, J. Wo¨rterbuch des Oskisch-Umbrischen (Heidelberg, 2000) Vetter, E. Handbuch der italischen Dialekte (Heidelberg, 1953) Vine, B. Studies in Archaic Latin Inscriptions (Innsbruck, 1993) Wackernagal, J. and A. Debrunner. Altindische Grammatik (Go¨ttingen, 1896
/1954) Walde, A. Lateinisches etymologisches Wo¨rterbuch . 3. Aufl. von J. B. Hofmann (Heidelberg, 1938) Warmington, E. Remains of Old Latin , vol. 1
/3 (London
/Cambridge, MA, 1936)

General Index adverbs in -per , 17
/21 ‘‘alacer’’ rule, 11n7 caesura Korschiana , 43
/44 diminutives, 64
/65, 75
/77 Homer, 7, 9, 43, 51, 78 Livius Andronicus, life and works, xi
/xiii

Osthoff ’s Law, 20, 54 ‘‘Sabine -l- ’’, 64 verbal morphology causative/iterative, 7
/8, 23, 29, 34 ‘‘double nasal,’’ 47
/50 -i' e=o- in Greek and Indo-Iranian, 8
/9, 37 ‘‘stative -e¯ -’’, 8
/9, 26
/29

meter, xiii
/xiv, 15, 43
/44 nominal morphology ablative sg., -d , 64 adjectives in -uus , 10, 77
/78 dative-ablative pl., -a¯bus , 43
/44 genitive sg., -a¯s, 23 i -stem substantives from o -stem adjectives, 10 o -stem adjectives from u -stem substantives, 10 verbal nouns in -a , 50
/52 vocative sg., -ie , 5
/6, 71

85

Index Verborum Albanian gjej, gje¨ndem, 48 Armenian gtane, 49
/50 harc‘ane, 49
/50 lizane, 49 lk‘ane, 49
/50 mah, marh, 10 Avestan afnahvant-, 57 (fra) manyeinte, 8, 38n4 sa¯ ra-, 78 French autre, 80 onze, 22n10, 56n32 prendre, 54 preˆt, 63 Saint-Sorlin, 76 Gothic aþþan, 56n23 (bi-, du-)ginnan, 47
/48 guma, 32 malan, 55n19 munþs, 41 sim(b)le , 19 ¯

Greek a ' delfea¤, 80 a ' deupia¤, 80 ' eiw, 78 ai‘mato ' omai, 65 a ' la a ‚jkriw, 11n16

a‚ krow, j

11n16 a ‚ mbrotow, 9 j a ' ntola¤, 51 aoidÆ, ¢ j 51 ' pestu ' w, 64 a ' r, 56n23 a ' ta ' Bromie, 6n2 gaye' v/ghye' v, 37 ga¤v, 37 daÆr, 64 ' w, 37 dolixo ßen, 19 j §paoidÆ, 51 ' w, 39 §ruyro ' kalxow, 80 ' mnv, 49 ka ' w, 80 kauxo ' niw, 35n24 ko kukke' vn, 27, 30n18 ' nv, 49 lagxa ' lambanv, 49 ' nv, 49 lanya le¤pv, 49 ' nv, 49
/50 limpa ' lu v, 39 ' nv, 49 manya ' omai, 41 masa me¤romai, 8
/11, 38n4 mo ˜i ra, 7
/11 MnÆmosunh, 23, 29 ' w, 67 nefro ' nu , nu˜ n, 20 ' mfh, 76 nu ßokriw, 11n16 j ' w, 36n25 o‘ malo ' o‘ mow, 36n25

87

88

Index Verborum ped¤on, 11n18 ' w, 11n18 pezo ' w/polio ' w, 39 pelio ' perusi, 53 ' w, 53 perusino ' w, 59n6 phro ph˜row, 59n6 p¤surew, 11n14 prodoxÆ, 51 proxe' v, 51
/52 proxoÆ, 51
/52 ' nomai, 49 punya ' w, 53 skai(W)o ' stellv, 62 ' low, 62 sto sunoxÆ, 51
/52 ' xa, 22n14 ta ' w, 22n14 taxu ' nv, 49 tugxa ' nv, 47
/50 xanda xe' v, 51
/2 xy–n, 32 xoÆ, 51
/52 j

h% uisu' ant-, 82n25 irmala-, 81n17 irmalant, 81n17 iyannai-, 13, 71 maklant-, 82n32 tekan, 32 Italian Bologna, 76 presto, 63 veleno, 76 Latin abavus, 52 -a¯ bus, 43
/44 amba¯ bus, 43 dextra¯ bus, 43
/44 dua¯ bus, 43 filia¯ bus, 43 aceo, 26 acidus, 27 acula, 76 ada¯ gium, 51 adeo, 65 adfatim, 10 adplu da, 55n20 ae nus, 56n31 ae s, 56n31 aestus, 10 ago, 77 alia, 53 (see also alia¯ s, under -a¯ s) alie nus, 53 alter, 80 amba¯ ge s, 55n17 ambo, see amba¯ bus, under -a¯ bus ambulo, 65
/66 ambustula¯ tus, 77 anteluculo, 81n12 aperio, 9
/10 applu da, 52 ¯

¯

Greek, Mycenaean pe-de-we-sa, 78 po-ri-wa, 39

¯

¯ ¯

Hittite assu-, 81n17 assuwant-, 81n17 daluka-, 37 h% ant, 57 h% appar, 57
/59 h% apparai-, 57
/59 h% appina-, 57
/59, 76 h% appinah% h% -, 57
/59 h% appinant-, 57
/59, 76 h% appinesˇ-, 57
/59 h% appir(iy)a-, 57
/59

¯

¯

¯

¯

¯

¯

¯

¯

Index Verborum

aqua, 76, 80 aquilus, 80 aquola, 76 arefacio, 29 areo, 29 arvum, 9 -a¯ s (genitive singular), 23 alia¯ s, 23, 53 esca¯ s, 23 familia¯ s, 23 fortuna¯ s, 23 Latona¯ s, 23 Mone ta¯ s, 23, 28
/29 terra¯ s, 23 via¯ s, 23 aspello, 55n20 asporto, 55n20 at, 52 atavus, 52 atque, 56n23 atquı¯, 56n23 audeo, 37
/38 autem, 64 autumo, 64 aveo, xv, 38n3 avidus, 37 Barcinona, 76 bel(u)va, 39 bibo, 75 bibulus, 74
/75 bı¯gae, 6 blandiloquentulus, 77 Bononia, 76 Bromie, 6n2 bucı¯tum, 29n14 buccula, 75 cado, 10, 77 caeduus, 82n20 calefacio, 29 caleo, 29 ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

89

Capitolinus, 76 Capitolium, 76 carnis (nom. sg.), 66n18 caro, 66n18 caupo, 63 caupona¯ , 63 cerno, 14, 16n14 cernulo, 82n22 cernulus, 78, 80 cernuus, 78, 80 certus, 14, 16n14 cinis, 35n24 clam, 64 clanculum, 64 commone facio, 29 compa¯ ge s, 55n17 condoce facio, 29 conta¯ gium, 55n17 convertuit, 41 corpusculus, 76 cre do, 75 cre dulus, 74
/75 cruentus, 74, 78 dedro, 16n9 dedron, 13, 16n9 de liculus, 77
/78 de licuus, 77
/78 de linquo, 77 de plumis, 68 de prans, 68 de puvio, 80 dexter, 43
/44 dingua, 64 doceo, 29 do, 13
/16, 71 dono, 14 duo, 43 e dentulus, 77 edo, 68 e mineo, 41 ¯ ¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯ ¯

¯

¯

¯

¯ ¯

¯

¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

90

Index Verborum

epula(e), epulum, xv, 58
/59 -e ta, -e tum, xiv, 23
/29 ace tum, 25
/27, 29 aescule tum, 23 arbore tum, 23 aspre tum, 25
/26 bu ce tum, 25
/26, 28 buxe tum, 24 castan(i)e tum, 24 coce tum, 25, 27 corne tum, 24 coryle tum, 24 cupresse tum, 23
/24 du me tum, 24, 28 fı¯ce tum, 24, 27
/28 fu ne tum, 25, 28 fime tum, 25 frutice tum, 24, 28 (h)arundine tum, 24 ¯ılice tum, 24, 28 iunce tum, 24, 28 ma¯ le tum, 24 mascule tum, 25, 28 Mone ta¯ s, 23, 28
/29 more tum, 25, 27 myrte ta, 24 myrte tum (mur-), 24 nuce tum, 24, 28 ole tum (‘excrement’), 25
/27 ole tum (‘olive-yard’), 24 olı¯ve ta, 25
/26 olı¯ve tum, 24, 26 olentice tum, 25, 27 palme tum, 24 pı¯ne tum, 24 pome tum, 24 popule tum, 24 porcule tum, 25, 26 querque tum (querc-), 24, 28 ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯

¯

¯ ¯

¯ ¯

¯

¯

¯

¯ ¯

¯

¯

¯

¯

rose tum, 24, 28 rube ta, 25
/29 rube tum, 24, 28 rumpotine tum, 24 sabule tum, 24, 26 salicte tum, 24 saxe tum, 24, 26 sentice tum, 24, 28 sepulcre tum, 24, 26 spı¯ne tum, 24, 28 te me tum, 25
/26 tucce tum, 25
/27 vı¯mine tum, 24 vı¯ne tum, 24, 28 vetere tum, 25
/26 virge tum, 24 -e tu do, 30n31 exiguus, 77 facul, 34 faecula, 75 fatuus, 10 fero, 61
/62 fı¯lia, 43 fı¯lius, 5
/6, 71 fla¯ vus, 39n1 foedifragus, 58 foedus, 58 fossa, 9 fraudulenter, 81n10 (in-)frendo, 67 frugis, 59n4 frustulum, 75 frux, 59n4 furvus, 39n1 gaudeo, 37
/38 gavı¯sı¯, 37
/38 gracente s, 75 gracilens, 75, 79, 81n9 gracilis, 74, 80 ¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

Index Verborum

91

gracilus, 74
/75 gracus, 75 GNAIVOD, 56n24 hemo, 32
/34 homo, xiv, 31
/35, 71 homon-, 31
/35, 71 homullus, 6, 35n2 horreo, 9, 38n4 iaculus, 78 incipio, 55n4 infans, 68 inmulgeo, 68
/69 inops, 59n3 -issimus, 33 istud, 21 iugum, 6 iungo, 6, 55n6 labrum, 68 lac, 69 Laertius, 5
/6 lanie na, 56n30 La¯ tı¯nus, 56n30, 76 La¯ tium, 56n30, 76 le vir, 64 lı¯bripens, 68 lingua, 64 luculenter, 81n10 luo, 39 lympha, 76 macellus, 75 macie s, 80 macilis, 81n13 macula, 81n13 mando, xv, 41
/42 maneo, 8, 38n4 manus, 43 medeor, 8 meminı¯, 23 Memoria, memoria, 23, 29n7 mentum, 41 ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

mereor, 7
/9, 38n4 mihi, 56n33 mil(u)us, 39 modero, 8 modestus, 8 modus, 8 mola, 52 moneo, 23, 28
/29, 34 mons, 41 morior, 9
/10 mors, 9
/10 Morta, 7
/11 mulleus, 39 mundus, 58 nancio, 53 ne, 34, 36n27 nec, 64 nefandus, 68 nefans, 68 nefa¯ s, 67 nefrens, 67
/68 nefrone s, 67 nefrundine s, 67 nego, 64 ne mo, 34, 67 nequeo, 15 nescio, 67 nex, 57 noceo, 77 -nont, -nunt, 13
/16 danunt, 13
/16, 71 exple nunt, 13
/16 ferı¯nunt, 13
/16 inserinuntur, 13
/16 nequı¯nont, 13
/16 obı¯nunt, 13, 15, 71 prodı¯nunt, 13, 15, 71 redı¯nunt, 13, 15, 71 (re -)novo, 66n15, 82n22 nudius tertius, 20 ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

92

Index Verborum

(ad-)nuo, 78 nullus, 36n26 nunc, 20 obsce nus, 53 obva¯ gula¯ tum, 77 occiduus, 10, 77 odor, 64 olea, 69 oleo, 26, 64 olidus, 27 olı¯va, 69 Opis, 59n4 opitulor, 58
/59 opitulus, 58
/59 ops, Ops, xv, 57
/59 opulens, 75, 81n9 opulenter, 75, 81n10 opulentus, 76 opus, xv, 58
/59 palleo, 39 pallidus, 39 pallor, 39 pa¯ lor, 65
/66 parum, 21n3 pascuus, 82n20 pavio, 80 pel(u)vis, 39 Pellaeo, 53 pendo, 68 -per, 17
/22 aliquantisper, 19 nuper, 18
/19 parumper, 18
/19 paul(l)isper, 18
/19 paul(l)umper, 18
/19 pauxillisper, 19
/20 quantisper, 19 semper, 18
/19 tantisper, 19 topper, 17
/22 ¯

¯

¯

¯

¯

¯

¯

¯

perspicuus, 10, 77 pestilens, 75, 80, 81n9 pestilita¯ s, 80 pestis, 80 peto, 77 petulans, 77 pisciculus, 76 plaudo, 52 pleo, 15 poc(u)lum, 61 pondus, 58 porcellus, 76 porcus, 75
/76 porculus, 75
/76 praebeo, 50 praebia, 55n16 praeda, xiv, 47, 50
/54 praehibeo, 55n16 praemium, 51 praendo, 53
/54 praeolo, 53 praesens, 63 praeses, 68 praesideo, 68 praesidium, 68 praesto (adv.), 61
/66 praesto (perf. praesta¯ vı¯), 66n15 praesto (perf. praestitı¯), 66n15 praestolo(r), xiv, 61
/66 praestu , 63
/64 praesulis, 80 praetor, 50
/51 praeustis, 53 praeut, 53 prandium, 68 prando, 68 (com-, de -) pre(he)ndo, 36n26, 47
/48, 52
/54 promptu, 63 pullus, 38
/39 ¯

¯

¯

¯

¯

¯

¯

¯

¯ ¯ ¯ ¯

¯

¯

¯

¯

¯

Index Verborum

quattuor, 10 quo (adv.), 65 quoad, 65 ra¯ vis, 19 ra¯ vus, 19 reguit, 41 re gula, 58 re iculus, 78 remora, 52 remoror, 52 rota, 52 rubeo, 26, 28, 36n29 ruber, 39 ru bidus, 27 rubus, 28 salio, 80 Sa¯ turnı¯nus, 76 scaevus, 53 se ce do, 52 secta, 9 se curus, 52 se dulo, 66n14 se dulus, 66n14 semel, 19 similis, 33 simul, 19, 33
/34 sino, 14
/16, 71 solino, 15, 16n6 solinunt, 13
/16 solvo, 39 specio, 10, 77 st, 33 stra¯ gulum, 58 sum, 33 taceo, 27 tam, 21 tego, 52 te mulenter, 81n10 terra, 56n31 terre nus, 56n31 ¯

¯ ¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

¯

93

testiculor, 77 testilor, 77 testis, 77 tı¯bı¯cen, 6 toga, 52, 54 tollo, 58 topper, see -per trux, 74 tum, 21 tunc, 22n12 turba, 76
/77 turbella, 75 turbo, 77 turbula, 75 turbulenter, 81n10 uber, 68
/69 -ulentus, xiv, 73
/82 aquilentus, 73, 76, 79
/80 ancunulentus, 73, 75, 79 bucculentus, 73, 75, 79 corpulentus, 73, 79 esculentus, 73, 79, 81 faeculentus, 73, 75, 79 fraudulentus, 73, 79 frustulentus, 73, 75, 79 gracilentus, 73, 78, 79
/81 iurulentus, 73, 79 lotiolentus, 73, 79 lu culentus, 74, 79, 81n12 lutulentus, 74, 79 macilentus, 74
/75, 79
/81 mustulentus, 74, 79, 81 obstru dulentus, 74
/75, 79 opulentus, 74
/76, 79 pestilentus, 74
/75, 79 pisculentus, 74, 79 posculentus, 74, 79, 81 pulverulentus, 74, 79 purulentus, 74, 79 rorulentus, 74, 79, 81 ¯

¯

¯

¯

¯

¯

¯

¯

¯

94

Index Verborum

sanguinolentus (-ul- ), 74, 79 te mulentus, 74, 79, 81 truculentus, 74, 79 turbulentus, 74
/75, 79 vı¯nolentus, 74, 79, 81 violentus, 74
/75, 77
/79 undecim, 22n10, 56n32 ustulo, 77 ustus, 77 va¯ gio, 77 valentulus, 77 vegeo, 8 vehis, 59n4 vene num, 76 vı¯cı¯nus, 53 vı¯cus, 53 video, 38 violens, 79, 81n9 violenter, 81n10 violo, 64, 74, 77 virgo, 32 vı¯s, 64 ¯

¯

¯

¯

¯

¯

¯

¯

Lithuanian die˜ vo, 56n24 einu`, 13, 71 mulvas, 39 pal˜ vas, 39 zˇmuo˜ , 32 Lycian epirijeti, 57
/58 xn˜ tawa-, 57 Middle Welsh genni, 47
/48 hynhaf, 33 mant, 41

Old English fealo, 39 (be-, for-)geitan, 47
/48 (bi-, on-)ginnan, 47
/48 guma, 32 mu ð, 41 rudu, 30n29 sı´m(b)les, 19 ¯

Old High German uoba, 58 uoben, 58 Old Irish aith-, 56n23 marb, 9 melid, 52 rethid, 52 ro-geinn, 47
/49 sinem, 33 tı´r, 56n31 Old Norse fo˛lr, 39 geta, 47
/48 muðr, 41 Old Church Slavonic mritvu% , 9 Oscan destrst, 33 eituas, 23 humuns, 31
/32, 34, 71 leginei, leginum, 6 meddı´ss, medı´keı´s, 31 pertiro-pert, 19 sakruvit, 5 su´m, 33

Index Verborum

statie, xiii tanginom, 6 u´´ıttiuf, 6n3 Sanskrit a´pas, 58 a¯´pas, xv, 58 apnas-, 57 a¯´pra-, 58 aprcchat, 49 ˚ a´´sri-, 11n16 a´vati, xv, 38n3 avidat, 49 bhanakti, 55n5 devar-, 64 deva¯´t, 56n24 hrsyati, 9, 38n4 ˚ jma´n, 32
/33 ks:ami, 33 mrta-, 9 ˚ nu, nu, 20 padvant-, 82n25 rudhira-, 39 smarate, 29n7 tapana-, 57 vindati, 49 yunakti, 55n6 ¯

95

South Picene me´fistru´i, 31 meitimu´m, 35n8 nemu´neı´, 31
/32, 34 safinu´s, 35n8 Spanish Barcelona, 76 Tocharian ka¯ tk-, 38n5 Umbrian amparitu, xiii heri, heris, xiii, 5 homonus, 31
/34, 71 maletu, 31 mehe, 56n33 natine, 6 tac¸ez, 27 tasetur, 31 triiu-per, 19 tutas, 23 sumul, 33
/34, 36n25 Vuvcis, 80 vutu, 80